Does obesity lower IQ? What the cohorts and genetic studies show

0 The short answer

Obesity is associated with lower IQ scores on average, but the best evidence points away from obesity lowering IQ and does not yet settle what produces the link. A 2024 systematic review and meta-analysis found school-age children with obesity scored 5.26 IQ points lower, yet a New Zealand birth cohort showed the gap present in childhood and no extra decline after people became obese, sibling comparisons shrank the link to almost nothing, and Mendelian randomization found no clear causal effect of body mass index on IQ. The reverse direction, lower childhood IQ before later obesity, is better supported in observational data, and in the British cohorts most of it is accounted for by education, which may itself reflect family background.

1 What Do the Studies Mean by Obesity and by IQ?

Each study on this page defines obesity by body mass index and measures IQ with a different test, so the first task is to say what is being compared before asking which causes which. The World Health Organization, in its fact sheet on obesity and overweight (read on October 20, 2026), defines obesity in adults as a body mass index of 30 or higher, and in children and adolescents aged 5 to 19 as a body mass index for age more than 2 standard deviations above the WHO growth reference median. The same page reports that 45 percent of adults, 2.6 billion people, were overweight in 2024, including 940 million, about 16 percent of all adults, living with obesity. It adds that other measures, such as waist circumference, can also identify overweight or obesity. Some of the genetic studies below use those other measures alongside body mass index.

Body mass index is a ratio of weight to height, and nearly every study reviewed here uses it or a percentile of it. That is a pragmatic choice that makes cohorts comparable, and it is also a limit. A study of body mass index tells us about weight relative to height, not about diet, activity, how long a person has lived with a higher weight, or why. When this page says obesity, it means the category defined by body mass index in that study, and the sections that follow name the definition each time it matters.

The IQ side is just as varied. The New Zealand cohort used the Wechsler Intelligence Scale for Children, Revised, in childhood and the Wechsler Adult Intelligence Scale, Fourth Edition, at age 38. The Aberdeen cohort used school-based tests of mental ability taken at ages 7 and 11. The Scottish cohort used the Moray House Test at age 11. The Israeli national study used an intelligence-equivalent score from a pre-military evaluation, and the genetic studies used reaction time, visual memory or composite cognitive scores from large genetic databases. These are different instruments sampling different abilities, which the page on what IQ measures and the page on the CHC model explain. A Full Scale IQ on a Wechsler test is a composite of several broad abilities, and a reaction time is a narrow speed measure, so a result for one is not automatically a result for the other.

Three kinds of claim need to be kept apart, and this page states each at the strength of the design behind it. An association says that two things go together in a group, and any study of a sample at one moment can show it. A direction claim says which came first, and it needs measurements at more than one time. A causal claim says that changing one would change the other, and it needs a design that rules out common causes: a randomized trial, which does not exist here because nobody can randomize obesity or IQ, or natural experiments such as sibling comparisons and genetic instruments.

One more point of language. A 2020 joint international consensus statement in Nature Medicine, written by a multidisciplinary group of international experts, reviewed evidence on weight stigma and concluded that it can cause physical and psychological harm and that people with obesity commonly face discrimination in workplaces, schools and healthcare settings. This page therefore writes about people with obesity, reports group averages as group averages, and gives no advice about weight or health, which is outside its scope and outside ACIS's competence. Neighboring questions are handled on other pages: the page on what can lower your IQ ranks exposures by the strength of evidence, the page on IQ and height covers body size and nutrition, and the page on exercise and IQ covers trials of physical activity.

2 Is Obesity Associated With Lower IQ? What the Two Meta-Analyses Found

Two systematic reviews, published 14 years apart, agree that obesity goes with lower IQ in children, and they disagree about which part of IQ differs and about what happens once education is held constant. The first, by Yu, Han, Cao and Guo in 2010 in Obesity Reviews, screened 883 citations from five databases and included 26 studies, most of medium quality. It reports a weighted mean difference in full IQ of minus 2.8 points between children with obesity and control children, and of minus 10.0 for performance IQ. The difference in verbal IQ, minus 7.01, and the difference in pre-school children, minus 15.1, were not statistically significant. The authors report that with increasing obesity the full IQ of pre-school children declined, with a significant difference for severely obese children. In pubertal children a slightly different pattern emerged.

The part of the 2010 review that matters most for direction concerns adults. Two studies reported an inverse association between full IQ and obesity in adults, but it was non-significant after adjusting for educational attainment. Four papers found that childhood full IQ was inversely associated with adult body mass index, which became null after adjusting for education. The authors concluded that, overall, there was an inverse association except in pre-school children, that after adjusting for education it was not significantly different from none, and that a lower childhood IQ was associated with obesity in later adulthood, perhaps with educational level mediating the persistence of obesity in later life.

The second review, by Yun and colleagues in 2024, also in Obesity Reviews, extended the search to studies published since June 22, 2009. According to its abstract, the systematic review covered 34 studies and the meta-analysis 17. It found a substantial correlation between obesity and reduced IQ, most notable among school-age children, with a mean difference of minus 5.26 points and a 95 percent confidence interval of minus 7.44 to minus 3.09. Within the verbal IQ subgroup the mean difference was minus 7.73 (95 percent confidence interval minus 14.70 to minus 0.77) in school-age children. The same paper ran a Mendelian randomization analysis, discussed in its own section below, and concluded that the genetic analysis implies a potentially weaker causal relationship and that future large cohort studies should address confounding.

Two features of these reviews deserve attention before any conclusion. The first is that they disagree on the verbal and performance split. The 2010 review found the larger significant gap on performance IQ and a nonsignificant gap on verbal IQ, while the 2024 review reported a significant verbal gap in school-age children. Verbal and performance IQ are Wechsler composites, not CHC abilities, and the pooled studies used different tests and ages, so neither review supports a claim that one broad ability is the one affected. The second is that a pooled mean difference across children is a statement about averages in cross-sectional and mixed designs. Both reviews call the relationship a correlation or association, and neither can say whether a child's weight preceded or followed the IQ score.

3 Every Study Side by Side

The design of each study fixes what it can say, so the table below sorts them by design, sample, measures and what they establish about direction. Figures are as the authors report them, and any conversion into points or shares is labeled as our arithmetic in the text that follows. The sections after the table take each design in turn.

Reading down the last column shows the pattern of the field. The designs that cannot fix direction, the cross-sectional ones, find a link. The designs that follow people from childhood find either that lower scores come before obesity or that there is no link, and the one that tested for a decline after obesity began found none. The designs that compare siblings or use genes find the link shrinking toward zero in one direction, and the genetic studies disagree about the other. The next sections go through them in that order.

4 Did Becoming Obese Lower IQ? What the Dunedin Cohort Tracked From Birth

The Dunedin study is the cleanest test of whether becoming obese lowers IQ, and it found no decline: the people who became obese already had lower scores in childhood. The paper by Belsky and colleagues in the American Journal of Epidemiology, published in 2013, analyzed the Dunedin Multidisciplinary Health and Development Study, a population-representative birth cohort of 1,037 children born in New Zealand in 1972 and 1973 and followed through their fourth decade with a 95 percent retention rate. Anthropometric measurements were taken at birth and at 12 later in-person assessments. The analysis sample was 913 people with the needed data. The authors framed the question directly: cross-sectional studies had found obesity associated with low intellectual ability, some had read that as obesity causing intellectual decline, and a prospective design was needed to test whether becoming obese was followed by a drop.

IQ was measured in childhood with the Wechsler Intelligence Scale for Children, Revised, averaged across ages 7, 9 and 11, and at age 38 with the Wechsler Adult Intelligence Scale, Fourth Edition. Obesity was defined as a body mass index at or above the 90th percentile of the United States reference distribution from ages 3 to 15, and as a body mass index of 30 or higher after age 15. The results, read from the paper, form a pattern in three parts. As expected, cohort members who became obese had lower adult IQ scores: about 96.93 against 101.28 for lean cohort members. But their childhood scores were lower by nearly the same amount, about 97.01 against 101.25, and the change from childhood to adulthood was close to zero for both groups, plus 0.04 for lean members and minus 0.08 for those who became obese, with no significant difference between them.

The authors checked whether the result held up under stricter tests, and it did. It was consistent when they accounted for birth weight and growth in the first years of life and when they looked only at childhood-onset obesity. It held for people whose obesity began after age 11, whose average score was about 96.4 in childhood and 96.6 in adulthood. It held among those with severe obesity, defined as obesity together with metabolic syndrome or elevated systemic inflammation, where the change was minus 0.23 points and not larger than in other groups. The study also ran a verbal memory test, a trail-making test and a pegboard motor task at ages 13 and 38. The group that became obese scored lower at both ages but showed no excess decline on those either. And the lower scores in children who later developed obesity were present as early as age 3, on a picture vocabulary test, where the average was about 97.80 for the later-obese group against 100.91 for the never-obese group.

The same paper also looked the other way. A lower childhood IQ predicted later obesity: the hazard of becoming obese rose by a factor of 1.28 (95 percent confidence interval 1.10 to 1.48) per standard deviation lower childhood IQ, and the relative risk of severe obesity was 1.40 (1.15 to 1.71). The authors' conclusion is a negative one about decline, not a positive one about cause in the other direction. They found no evidence that obesity contributed to a decline in IQ, and the design supplies the temporal order that the cross-sectional studies lacked. They do not claim that low IQ causes obesity. A lower score at 3 could reflect family circumstances, early development or anything else that also raises the chance of later obesity, and the study cannot separate those.

Two limits should be stated plainly. The cohort is one cohort in one country, born in two years, so the pattern may not carry to other places and generations. And the finding concerns the first half of life, from childhood to the late thirties, so it speaks to whether obesity lowers IQ in young and middle adulthood and not to later-life cognitive decline, which is a different literature and outside this page.

5 Does a Lower Childhood IQ Predict Obesity, and What Explains the Link?

Cohorts that began in childhood find that lower IQ goes with later obesity, and in the British cohorts most of that link is accounted for by education, which may itself reflect family background. The reversed question, whether an early score predicts later weight, has been asked in several large cohorts, and the answers are consistent in direction and divided in interpretation.

The British 1958 birth cohort is the largest. Chandola, Deary, Blane and Batty analyzed the National Child Development Study, a prospective cohort of 17,414 births in Great Britain in the late 1950s, with IQ at age 11 and body mass index at 16, 23, 33 and 42. In unadjusted analyses a lower childhood IQ went with a higher prevalence of obesity at 42. The odds ratio per standard deviation lower IQ was 1.38 (95 percent confidence interval 1.26 to 1.50) in women and 1.26 (1.15 to 1.38) in men. The association stayed significant after adjusting for childhood characteristics such as socioeconomic factors, and was heavily attenuated after control for adult characteristics, particularly education: 1.11 (0.99 to 1.25) in women and 1.10 (0.98 to 1.23) in men. For weight gain between ages 16 and 42, the authors report that education and dietary characteristics in adult life mediated the association with childhood IQ. Their conclusion is that the relation appears to be largely mediated through educational attainment and the adoption of healthy diets in later life.

The Aberdeen cohort of children born between 1950 and 1956 shows the same attenuation. Lawlor, Clark, Davey Smith and Leon followed 5,467 people from childhood to a 2001 survey, with intelligence at age 7 and adult body mass index at a mean age of 48. Each standard deviation of childhood intelligence went with a body mass index lower by 0.35 kilograms per square meter (95 percent confidence interval minus 0.49 to minus 0.21), and each step on a seven-category education scale went with 0.28 lower. After adjusting for education, the intelligence estimate fell to minus 0.03 (minus 0.19 to 0.13), essentially none, while the education estimate was not affected by adjusting for intelligence. A second paper on the same cohort, by Batty, Deary and Macintyre in 2007, analyzed 5,340 people and found that each standard deviation higher IQ at age 11, about 12.97 points, went with an odds ratio of 0.78 (95 percent confidence interval 0.72 to 0.83) for self-reported obesity in middle age, sex adjusted. According to that paper the obesity association remained statistically significant after adjusting for childhood and adult socioeconomic position, education and birth weight. The two Aberdeen papers differ in outcome (continuous body mass index against obesity), in age at testing (7 against 11) and in measurement (obesity was self-reported in the second). No reconciliation of the two appears in the papers we opened, so the honest summary is that the same cohort gives a near-null result for body mass index after education and a persisting association for obesity.

The Scottish 1936 birth cohort gives no association at all, with a caveat about size. Harris, Brett, Deary and Starr followed the 36-Day Sample of the 1947 Scottish Mental Survey, which tested 6,291 children at age 11. Only 423 people, about 6.7 percent of the original sample, were measured again in late adulthood, at about 78. The authors report no significant relationship between body mass index and intelligence in childhood or in late adulthood, and no longitudinal link: childhood intelligence did not predict late-life body mass index (a standardized coefficient of minus .04, p of .44). They suggest the null result may partly reflect the small, selected follow-up sample, which is a fair reading, because the people who return decades later are not a random subset.

Against these sits one paper that disagrees on the central point. Kanazawa, in a 2013 paper in Obesity, used the same British 1958 cohort, with childhood intelligence at 7, 11 and 16 and body mass index at 51 for 17,419 people. He reports that childhood general intelligence has a direct effect on adult body mass index, obesity and weight gain, net of education, earnings, parental body mass index, childhood social class and sex, and that more intelligent children grew up to eat more healthy foods and exercise more frequently as adults. The abstract opens by saying that recent studies conclude there is no direct effect net of education but that evolutionary psychological theories suggest otherwise. It is a single-author analysis of the same cohort with a later age at outcome (51 against 42), intelligence drawn from three ages instead of one, and a different set of adjustments. It is not a replication in a new sample, so it is best read as a dissent that needs independent confirmation.

Why does adjusting for education matter so much? There are two readings, and the data alone do not choose between them. In the first, education is a pathway: higher childhood IQ leads to more schooling, which leads to different jobs, incomes, diets and health habits, which lead to different weight. Chandola and colleagues describe the link as largely mediated by education and diet, and Lawlor and colleagues conclude that education explains the intelligence association. In that reading, controlling for education removes the part of IQ's association that passes through it, so a null after adjustment does not mean IQ does nothing. In the second, education is a marker of family background. Children from more advantaged homes score higher, stay in school longer and live in settings with different food and exercise environments, and IQ and education both reflect that background. The sibling analyses in the next section address the second reading.

6 Do the Differences Survive a Comparison Between Siblings?

When brothers and sisters from the same family are compared, the link between early cognitive ability and adult body mass index shrinks to almost nothing, which points to shared family background rather than to ability itself. Siblings share parents, neighborhood, household income and much of their genetic makeup, so a difference within a pair is less likely to reflect those shared factors. It is not a randomized experiment, because siblings differ in ways that are not shared, but it removes a large block of confounding.

The Aberdeen paper by Lawlor and colleagues included a sibling analysis of 643 pairs, 1,286 individuals, and it did this for education, not for intelligence. The within-pair estimate of education on adult body mass index was minus 0.06 (95 percent confidence interval minus 0.26 to 0.14), while the between-family estimate was minus 0.37 (minus 0.58 to minus 0.17), a difference with a p value of 0.03. The authors read this as suggesting that fixed family and neighborhood factors may play a role in the link between education and body mass index.

The direct test for cognitive ability is the 2023 study by Wright, Davies and Bann in PLOS Medicine. It pooled four American cohorts, the National Longitudinal Survey of Youth 1979, its Children and Young Adults sample, the 1997 survey, and the Wisconsin Longitudinal Study, giving 12,250 full siblings from 5,602 households and 118,355 observations. Cognitive ability was measured in adolescence and body mass index in adulthood, from ages 20 to 64. Between families, moving from the 25th to the 75th percentile of cognitive ability went with a body mass index lower by 0.95 kilograms per square meter (95 percent confidence interval minus 1.21 to minus 0.69), and adjusting for family socioeconomic position reduced that to minus 0.61 (minus 0.90 to minus 0.33). Within families the estimate was minus 0.06 (minus 0.35 to 0.23), with an interval that spans zero.

The authors conclude that the association between cognitive ability and body mass index, which is well established between families, may largely reflect confounding by family background factors. They add that the small within-family estimate leaves room for a limited causal role that the data cannot establish. They also list limitations: height and weight were self-reported, non-shared confounders could remain, one sibling's outcomes can influence another's, and all the data come from one country. For scale, a body mass index difference of 0.06 kilograms per square meter is, by our arithmetic, about 0.17 kilograms of body weight for a person 1.7 meters tall.

How does this fit with Dunedin? Dunedin shows that the gap between those who go on to develop obesity and those who do not is already present very early, and it cannot say why. The sibling study suggests that much of the population-level association is shared with family circumstances, which would be one answer. Neither says that nothing about the child contributes, and both leave open the weaker claim that a small independent effect exists.

7 What Does the Genetic Evidence Say?

Mendelian randomization studies find no clear causal effect of body mass index on IQ, and the evidence on the reverse direction is mixed, so genetics does not support the claim that higher weight lowers intelligence. The method uses genetic variants that influence a trait as stand-ins for the trait, on the logic that variants are assigned at conception and so are not changed by later circumstances. A guide for clinicians by Davies, Holmes and Davey Smith in the BMJ describes it as using genetic variation as a natural experiment to investigate causal relations in observational data, and stresses that findings depend on specific assumptions. In the usual framework, as we summarize it, the variants must be associated with the exposure, must not be associated with confounders, and must affect the outcome only through the exposure. A violation of any one can bias the result, which is why the estimates are read alongside other designs.

The 2024 paper by Yun and colleagues is the Mendelian randomization result closest to the title question. It relied on genome-wide association databases and the Genetic Investigation of Anthropometric Traits consortium, and found no significant causal relationship between body mass index and IQ, either in childhood or in adulthood. Its authors read the whole analysis as showing a significant correlation but a potentially weaker causal relationship.

The largest individual-level study is Norris, Salzmann, Henry, Garfield and Pinto Pereira in the International Journal of Epidemiology, 2023, which used 378,877 UK Biobank participants of European ancestry aged 40 to 69. It tested body mass index, body fat percentage and waist-hip ratio against reaction time and visual memory, in both directions, with 76 genetic variants for body mass index. For adiposity as the exposure, the estimates were imprecise, mostly consistent with the observational direction, and with confidence intervals that generally included zero. The authors conclude that the observational association of adiposity with cognitive function is likely not causal. For the reverse direction the result was not what a simple story of low ability leading to obesity would predict. The authors report that worse genetically predicted visual memory was associated with lower adiposity, with an inverse-variance weighted estimate of minus 3.57 percent for body mass index (95 percent confidence interval minus 5.15 to minus 1.84), while reaction time showed no clear effect. They say findings should be triangulated with other cognitive outcomes and methods, and these two measures are narrow ones compared with a full IQ test.

Not every genetic analysis agrees. Gong and colleagues in Frontiers in Endocrinology, 2025, ran a bidirectional Mendelian randomization with summary data and reported that higher genetically predicted body mass index, body fat percentage and waist-hip ratio were associated with lower cognitive ability measures, and that worse fluid intelligence and simple reaction time were associated with higher body mass index and lower HDL cholesterol respectively. The abstract does not report effect sizes or sample sizes, so a reader cannot weigh it against the larger individual-level study, and it offers no way to tell whether the difference from Norris and colleagues comes from the cognitive measures, the data or the methods.

Why does a family design matter for genetics too? A methods paper by Brumpton and colleagues in Nature Communications, 2020, showed that estimates from unrelated people can be biased by dynastic effects, assortative mating and population stratification. Using 61,008 siblings, it found that genetic estimates from unrelated individuals suggested that lower body mass index increases educational attainment, but these effects were strongly attenuated in within-family analyses, while the effect of lower body mass index on diabetes and high blood pressure held in both. A later within-sibship study by Howe and colleagues in the International Journal of Epidemiology, 2023, using 72,932 siblings, found that both population and within-sibship estimates supported educational attainment decreasing body mass index, and that the two were largely consistent. Neither paper studied IQ, and the Yun abstract does not say whether its instruments were checked within families, so the lesson is about method: for traits tied to education and family background, a genetic estimate from unrelated people deserves a within-family check before it is read as causal.

Taken together, the genetic evidence gives no robust support for a causal effect of higher body mass index on IQ, partial and conflicting support for an effect of some cognitive measures on adiposity, running in opposite directions in the two analyses that tested it, and an unresolved disagreement between them. That is a weaker statement than either side of the popular debate makes, and it is the one the studies allow. The page on whether IQ is genetic explains why heritability or a genetic correlation is not the same as a causal effect of one trait on another.

8 What Did 2.48 Million Israeli Adolescents Add, and What Could They Not?

The largest study of body mass index and cognitive performance is a national snapshot of adolescents, and its scale makes the association precise without making it causal. Simchoni and colleagues, writing in the European Journal of Endocrinology in 2023, analyzed Israeli-born adolescents aged 16 to 20 who underwent a pre-military evaluation between 1967 and 2018, about 2.48 million people. Body mass index came from measured height and weight, and cognitive performance from a validated intelligence-equivalent test standardized by year and sex. Among males with severe obesity, 29.4 percent scored below the 25th percentile, against 17.7 percent of normal-weight males. For about 445,000 participants, parental cognitive scores were also available, and the estimates held after adjusting for sociodemographic factors, coexisting conditions and those parental scores.

Two details guard against a simple reading. The relationship in males was J-shaped, not a straight line: compared with normal weight, the odds of a low cognitive score were 1.45 for underweight, 1.13 for overweight, 1.36 for mild obesity and 1.58 for severe obesity, and females showed similar patterns. A higher chance of a low score at the underweight end as well as the obesity end suggests that something other than adiposity as such, such as health, nutrition or family circumstances, is tied to both. And the design is cross-sectional, one measurement in adolescence, so it cannot say which came first. The authors conclude that obesity is associated with increased odds of a lower cognitive performance, which is a statement about association. The study adds precision and the J shape, and nothing about direction.

9 Why Might Obesity and IQ Be Linked at All?

The proposed explanations differ in how much evidence each has, and the studies above test only some of them. Ranked by what the designs can support, they are these. Family and neighborhood background is the best supported: the association shrinks within siblings (Wright and colleagues), and the education link shrinks within Aberdeen sibling pairs. Education and adult diet as a pathway is supported for the British cohort, where they mediated the association with childhood IQ, although the two readings of education in the earlier section cannot be separated by those data. Early development is plausible because the Dunedin gap is present at age 3, but the study does not identify what produces it.

Obesity damaging the brain, through inflammation or metabolic disturbance, is the explanation the question in the title implies, and it is the one the cohorts do not support in young and middle adulthood. In Dunedin there was no excess decline even among people with metabolic syndrome or elevated inflammation, and the genetic studies found no clear causal effect of body mass index on IQ. A narrower version, that obesity affects specific functions such as memory or executive control, has been studied, and the Dunedin tests of those functions also showed no excess decline, though that is one cohort. Low IQ causing obesity directly, through choices about food and activity, is supported by Kanazawa's single analysis and not supported by the sibling and genetic designs, so it remains unproven.

Finally, measurement matters. IQ and body mass index are each measured with error, and studies of children, adolescents and adults use different tests. A real but small link can appear or vanish with the instrument, and that may be one reason the verbal and performance split differs between the two meta-analyses. The page on reliability and validity explains why a score carries uncertainty, and the page on how IQ scores are normed explains what a score is compared with.

10 How Big Is a Gap of Four or Five IQ Points?

A gap of four or five IQ points between groups is about a third of a standard deviation, which is real as an average and small against the spread of individuals. The Dunedin groups differed by about 4.2 points in childhood, our arithmetic on the reported means of 101.25 and 97.01, and the 2024 meta-analysis by 5.26 in school-age children. On a scale with a standard deviation of 15, these convert to roughly 0.28 and 0.35 of a standard deviation. If two groups are normally distributed with equal spread and differ by 0.35 of a standard deviation, about 64 percent of the higher group scores above the average of the lower group, and a person drawn at random from the higher group outscores a person drawn at random from the lower group about 60 percent of the time. Put another way, the average of the lower group would sit near the 36th percentile of the higher group.

That overlap is why group differences cannot be turned into statements about a person. The page on IQ score versus percentile shows how scores map to ranks, and the percentile calculator turns any score into one. A gap of five points is small next to the spread of scores within either group, which runs across most of the scale, and the IQ score chart shows how wide that scale is. The Israeli figures make the same point differently: 29.4 percent of males with severe obesity scored below the 25th percentile, which means about 71 percent did not, by our arithmetic.

11 What This Does Not Say About You, Your Child or Anyone With Obesity

A group difference in average scores, a cohort pattern and a null genetic result say nothing about the ability of any particular person, and they do not say that weight and intelligence are related in your case. The studies compare averages and shares of groups, and the people inside each group are spread across the whole range of scores. A person with obesity is not likely to score lower or higher because of it, and a lower score is not a sign that a person will develop obesity. None of these studies identifies an individual at risk, and none supports using weight to estimate ability or ability to estimate weight.

For a parent, the evidence does not support concern that a child's weight is lowering the child's intelligence, because the strongest cohort found no excess decline after obesity began. For an adult, the evidence does not say that a score reflects weight or that weight loss or gain would change it, since no trial here tested that. Weight stigma, which the 2020 consensus statement describes as harmful, is not supported by these results either. And any question about health, nutrition or treatment belongs with a clinician, not with an IQ page. The way to learn what a person can do is to measure it directly with a standardized test and read the result with its confidence interval. The page on what IQ scores mean explains how, the page on online IQ tests explains what a normed online test can and cannot tell a person, and ACIS is an online, unsupervised test that is not a clinical or diagnostic instrument. The sibling pages on empathy and IQ, math ability and IQ, deliberate practice versus IQ and high-IQ quirks apply the same discipline to other claims.

12 What the Evidence Supports, Stated Narrowly

The evidence supports four narrow statements about obesity and IQ, and each stops short of the claim that obesity lowers intelligence or that low intelligence causes obesity. First, groups with obesity score lower on IQ tests on average, by about 5 points in school-age children in the 2024 meta-analysis and by a few points in the best cohorts, and the national Israeli study confirms the association at scale. Second, in the one cohort that measured IQ before and after, becoming obese was not followed by an excess decline, and the gap was present in childhood. Third, the link between childhood intelligence and later weight is largely accounted for by education for body mass index and weight gain in two British cohorts, and between siblings it shrinks to almost nothing. Fourth, Mendelian randomization finds no clear causal effect of body mass index on IQ, while the evidence for the reverse direction is mixed and the analyses disagree.

What the evidence does not support is any claim that obesity damages intelligence in early or middle life, any claim that lower intelligence directly causes obesity, any use of weight to estimate ability, and any inference to an individual. The designs are observational or genetic, the measures differ, and one of the most cited disagreements, Kanazawa against the mediation analyses, rests on the same cohort. The Standards for Educational and Psychological Testing (AERA, APA and NCME, 2014) define validity as the degree to which evidence and theory support the interpretations of test scores for proposed uses. By that standard, reading a person's weight as a sign of their ability, or their score as a statement about their weight, is an interpretation the evidence does not support. The American Psychological Association's Ethical Principles of Psychologists and Code of Conduct, in the version effective January 1, 2017, addresses interpretation in Standard 9.06: psychologists take into account the purpose of the assessment and test factors, test-taking abilities and other characteristics of the person assessed, such as situational, personal, linguistic and cultural differences, that might reduce the accuracy of their interpretations, and they indicate any significant limitations. A group result from a cohort is not an individual interpretation of that kind.

13 Sources Behind This Page

Every figure above is traceable to one of the following, and each is linked at the point where it is used. Sample sizes, differences, odds ratios and intervals are quoted as the authors report them, and any conversion into standard deviations, overlap or body weight is labeled as our arithmetic. Journal records were confirmed with Crossref and PubMed, the abstracts were read through PubMed or the publisher or repository record, and the Dunedin, Scottish, American sibling and UK Biobank papers were opened on PubMed Central on October 20, 2026. The World Health Organization page was read on October 20, 2026. Publisher pages that block automated readers are cited by DOI after the record was confirmed.

• Yun S Y, Yun J Y, Lim C, Oh H, Son E, Shin K, Kim K, Ko D S and Kim Y H. Exploring the complex link between obesity and intelligence: Evidence from systematic review, updated meta-analysis, and Mendelian randomization. Obesity Reviews, 2024, volume 25, issue 12, article e13827.
• Yu Z B, Han S P, Cao X G and Guo X R. Intelligence in relation to obesity: a systematic review and meta-analysis. Obesity Reviews, 2010, volume 11, issue 9, pages 656 to 670.
• Belsky D W, Caspi A, Goldman-Mellor S, Meier M H, Ramrakha S, Poulton R and Moffitt T E. Is obesity associated with a decline in intelligence quotient during the first half of the life course? American Journal of Epidemiology, 2013, volume 178, issue 9, pages 1461 to 1468.
• Chandola T, Deary I J, Blane D and Batty G D. Childhood IQ in relation to obesity and weight gain in adult life: the National Child Development (1958) Study. International Journal of Obesity, 2006, volume 30, issue 9, pages 1422 to 1432.
• Kanazawa S. Childhood intelligence and adult obesity. Obesity, 2013, volume 21, issue 3, pages 434 to 440.
• Lawlor D A, Clark H, Davey Smith G and Leon D A. Childhood intelligence, educational attainment and adult body mass index: findings from a prospective cohort and within sibling-pairs analysis. International Journal of Obesity, 2006, volume 30, issue 12, pages 1758 to 1765.
• Batty G D, Deary I J and Macintyre S. Childhood IQ in relation to risk factors for premature mortality in middle-aged persons: the Aberdeen Children of the 1950s study. Journal of Epidemiology and Community Health, 2007, volume 61, issue 3, pages 241 to 247.
• Harris M A, Brett C E, Deary I J and Starr J M. Associations among height, body mass index and intelligence from age 11 to age 78 years. BMC Geriatrics, 2016, volume 16, article 167.
• Wright L, Davies N M and Bann D. The association between cognitive ability and body mass index: A sibling-comparison analysis in four longitudinal studies. PLOS Medicine, 2023, volume 20, issue 4, article e1004207.
• Simchoni M, Derazne E, Pinhas-Hamiel O, Cukierman-Yaffe T, Bendor C D, Bardugo A, Chodick G, Tzur D, Endevelt R, Gerstein H C, Afek A and Twig G. Adolescent body mass index and cognitive performance: a nationwide study of 2.48 million Israeli adolescents. European Journal of Endocrinology, 2023, volume 189, issue 1, pages 67 to 77.
• Norris T, Salzmann A, Henry A, Garfield V and Pinto Pereira S M. The relationship between adiposity and cognitive function: a bidirectional Mendelian randomization study in UK Biobank. International Journal of Epidemiology, 2023, volume 52, issue 4, pages 1074 to 1085.
• Gong M, Liu H, Liu Z, Wang Y, Qi S, Guo H and Jin S. Causal links between obesity, lipids, adipokines, and cognition: a bidirectional Mendelian-randomization analysis. Frontiers in Endocrinology, 2025, volume 16, article 1439341.
• Brumpton B, Sanderson E, Heilbron K and colleagues. Avoiding dynastic, assortative mating, and population stratification biases in Mendelian randomization through within-family analyses. Nature Communications, 2020, volume 11, article 3519; and Howe L J, Rasheed H, Jones P R and colleagues. Educational attainment, health outcomes and mortality: a within-sibship Mendelian randomization study. International Journal of Epidemiology, 2023, volume 52, issue 5, pages 1579 to 1591.
• Davies N M, Holmes M V and Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ, 2018, volume 362, article k601.
• World Health Organization. Obesity and overweight, fact sheet, read October 20, 2026; and Rubino F, Puhl R M, Cummings D E and colleagues. Joint international consensus statement for ending stigma of obesity, Nature Medicine, 2020, volume 26, issue 4, pages 485 to 497.
• American Educational Research Association, American Psychological Association and National Council on Measurement in Education. Standards for Educational and Psychological Testing, 2014; and American Psychological Association, Ethical Principles of Psychologists and Code of Conduct, Standard 9.06, version effective January 1, 2017.

14 Frequently Asked Questions