Stress, trauma and IQ: what the studies show about measured ability

0 The short answer

Severe, prolonged institutional deprivation in early childhood is followed by markedly lower measured IQ, and a randomized trial shows part of that gap can be closed, but the stress and maltreatment found in ordinary families show much smaller associations once earlier ability and family background are taken into account. In a randomized trial of Romanian institutions, children placed in foster care scored about 9 points higher at age 18 than children randomized to stay in institutional care, yet young adults with any institutional history still averaged about 26 points below community peers. Birth cohorts with an early IQ baseline found the link between victimization and later IQ shrank to near zero after adjustment, a smaller low income cohort that adjusted for maternal IQ still found about half a standard deviation, and acute laboratory stress impairs working memory without clearly changing total IQ. No study can split one person's score into a state part and a trait part.

1 What Counts as Stress, and What Counts as Childhood Adversity?

The word stress names at least five different exposures in the IQ literature, and a number measured for one of them cannot be carried over to another. Acute stress is a short physiological response to a threat, usually induced in a laboratory for a few minutes. Chronic stress is a demand that persists for months or years, such as job strain or constant worry. Toxic stress is the pediatric term for strong, frequent and prolonged adversity in childhood. Maltreatment and violence cover abuse, neglect, bullying and domestic violence. Institutional deprivation means being raised without consistent, responsive caregiving.

The Center on the Developing Child at Harvard University, on a page read on October 6, 2026, separates positive, tolerable and toxic stress. Shonkoff, Garner and colleagues set out the underlying framework in a 2012 report in Pediatrics. A second framework, by McLaughlin, Sheridan and Lambert, divides adversity by what it takes away. Deprivation is the absence of expected environmental input and complexity, and threat is experience that endangers physical integrity. The authors argue that the two shape development differently and cannot be reduced to stress pathways. Because institutional rearing is mostly deprivation, orphanage evidence informs the stress question by analogy, not directly.

Measured IQ here means a score on a standardized test such as a Wechsler scale, with a mean of 100 and a standard deviation of 15, as the page on how IQ scores are normed explains. Many stress studies measure narrower abilities, usually working memory. The page on what IQ measures and the page on the CHC model show how those abilities sit inside a full battery.

Four neighboring questions belong elsewhere. The page on anxiety and IQ covers test anxiety, the page on poverty and IQ covers income, the page on environmental toxins and IQ covers pollutants such as lead, and the page on sleep and IQ covers the night before a test. This page offers no diagnostic reading of any person and no advice about coping.

2 State or Trait: Two Questions Hidden in One

A score can be low because of how a person was doing on the day or because of how their abilities developed over years, and no study of stress separates the two for an individual. A state effect is a temporary change in performance caused by the person's condition at testing, and it reverses when the condition does. A trait or developmental effect is a lasting difference in the underlying ability. The page on sleep and IQ draws the same line for sleep loss.

The two readings lead to opposite conclusions about a low score. A state effect means it understates what the person can usually do, and a developmental effect means it describes the ability built so far. Reliability does not settle which applies. As the page on reliability and validity explains, reliability concerns how stable a score is across occasions, while validity concerns whether an interpretation is supported.

Different designs answer different halves of the question. Laboratory experiments and natural experiments tied to the timing of a test speak to state effects. A randomized trial that changes a child's environment for years speaks to developmental effects. Birth cohorts speak to them only if earlier ability and family background are measured and counted. Cross-sectional surveys cannot separate state, trait or direction.

One piece of evidence bears on timing. Matte-Landry and colleagues pooled longitudinal studies of complex trauma in 2023 and found poorer overall cognition than controls when the trauma was recent, and similar cognition when it was less recent. The authors speculate that transient effects fade, and they note that only 2 of the 13 studies reviewed measured cognition both before and after the trauma. That hints that part of the gap is a state without proving it.

3 Every Design Side by Side

Read by design rather than by headline, the estimates run from near zero in adjusted cohorts and laboratory tests to 9 points in the one randomized trial, while unadjusted gaps between institution and family reared children reach 20 points and more. The first table covers exposures that last years. The second covers acute stress and chronic stress in adolescents and adults. Results are quoted as the authors report them, and the cell says where a figure is our subtraction or our arithmetic.

4 The Randomized Trial: The Bucharest Early Intervention Project

The Bucharest Early Intervention Project is the strongest causal design in this literature, and it found a lasting IQ gain of about 9 points from foster care alongside a large remaining gap to community peers. The project enrolled children abandoned at or shortly after birth and living in institutions in Bucharest, Romania. According to the 2023 multilevel analysis by King and colleagues, 136 children aged about 6 to 31 months were randomized in equal numbers to foster care or to continued institutional care by drawing slips from a hat. Foster care was extremely limited in Bucharest, so the investigators built a network of 56 families. The authors describe the project as the only randomized trial of foster care as an alternative to institutional care. Nelson and colleagues reported in Science in 2007 that, through 54 months, children who remained in institutions scored markedly below never institutionalized children and below the foster care group. The intervention aimed to improve the relationship between caregiver and child, not to raise IQ.

Ages 8, 12 and 18

At age 8, Fox and colleagues tested 103 children with the Wechsler Intelligence Scale for Children, fourth edition (WISC-IV). Full scale IQ averaged 81.5 with foster care, 76.2 with care as usual and 107.0 in 38 community children. The 5.3 point difference is our subtraction, and the authors call it marginal, with a p value of .07, although the verbal comprehension difference was significant. At age 12, Almas and colleagues tested 107 children. Averages were 75.8, 68.8 and 98.6, and the intent to treat difference of about 7 points (our subtraction) was significant. Of the 52 care as usual and 53 foster care children with IQ data, 30 and 18 scored below 70, and 10 more children were too impaired to be tested.

At age 18, Humphreys and colleagues tested 95 trial participants and 40 community young adults. Foster care averaged 73.5 and care as usual 64.7, a difference of 9.00 points (p of .026), which the abstract states as 0.6 standard deviations. The 95 participants with any institutional history averaged 69.2 against 94.8 for the community group, a gap of 26.2 points that the authors place in the very to extremely low range. They conclude that early high quality care did not fully remediate the effects of deprivation. The analysis used a model designed for floor effects, and the community group was a comparison, not a matched control.

Does the Gain Fade?

King and colleagues pooled 7,088 observations from 130 children across all waves, covering IQ, growth, brain activity and psychopathology. IQ, standardized within each wave, was higher with foster care by 0.39 standard deviations, with a 95 percent interval from 0.16 to 0.63. Age of assessment did not significantly change the pooled effect on cognitive, physical and neural outcomes, and the authors describe the benefits as sustained through late adolescence. All but 13 of the care as usual children had entered family care by 18, so the randomized contrast is early foster care against mostly later family care, an intent to treat analysis that Fox and colleagues call conservative.

Humphreys and colleagues report that caregiving quality statistically explained 49 percent of the intervention effect and attachment security 71 percent. Those are mediation estimates from a trial not built to separate mechanisms, and the authors say nutrition, school and neighborhood cannot be ruled out. Stable placements were associated with larger gains, but stability was not randomized, and children with lower IQ may be at greater risk of placement disruption.

What the Trial Cannot Say

The trial shows that environment can change IQ in children raised in severe deprivation. It does not test whether everyday stress lowers IQ. Institutional rearing removes consistent, responsive caregiving, which fits the deprivation dimension that McLaughlin and colleagues distinguish from threat. The sample is small, the 18 year follow-up lost participants, and the foster care was supported by a research team.

5 Adoption Natural Experiments: Romanian Adoptees and the Orphanage Meta-Analyses

Adoption studies show that children raised in institutions begin far behind and recover part of the gap in family homes, but how much they recover depends on how long the deprivation lasted and on which measure is used. These are natural experiments, so selection into adoption is a standing concern.

The English and Romanian Adoptees Study

The English and Romanian Adoptees (ERA) study followed children who spent from soon after birth to as long as 43 months in severe deprivation in Romanian institutions before adoption into UK families. Rutter and the ERA team compared, at age 4, 111 Romanian children who reached the UK before age 2 with 52 UK adoptees placed before 6 months. Catch-up was nearly complete for those who arrived before 6 months and impressive but incomplete for those who arrived later. The abstract reports a mean McCarthy General Cognitive Index of 92 against 109 for the within UK adoptees. At age 11, Beckett and colleagues compared 131 Romanian adoptees with 50 UK adoptees. They report marked persisting effects for many children over 6 months old on arrival, a 15 point decrease for that group, and no differentiation within the range of 6 to 42 months.

The young adult follow-up defined cognitive impairment as an IQ below 80. In the group with more than 6 months of deprivation, 98 were assessed at age 6 and 72 in young adulthood, against 52 and 39 UK controls. Impairment rates were markedly higher than in controls at 6 and 11 and back to normal in young adulthood, while symptoms of autism spectrum disorder, disinhibited social engagement and inattention stayed elevated. A brain imaging study of the same cohort by Mackes and colleagues scanned 67 adoptees, deprived for 3 to 41 months, and 21 UK adoptees. Total brain volume was 8.6 percent smaller and related to the duration of deprivation, and the authors report that it statistically mediated the links to lower IQ and to attention and hyperactivity symptoms. The abstracts give no adult mean IQs, so these findings cannot be compared directly with the Bucharest means.

Orphanage and Adoption Meta-Analyses

Van IJzendoorn, Luijk and Juffer pooled 75 studies of 3,888 children in 19 countries. Children in children's homes had a mean IQ of 84, against 104 in families, a 20 point gap that the authors report as a trimmed standardized difference of 0.74. The delay was associated with age at placement, age at assessment and the development level of the country. These observational comparisons describe how large the gap tends to be, not what one institution does to one child. Van IJzendoorn, Juffer and Poelhuis pooled 62 studies of 17,767 adopted children and found them scoring higher than siblings or peers who stayed behind and no different in IQ from nonadopted peers in the adoptive families, with school performance and language lagging; the abstract gives no IQ advantage in points.

What Recovers and What Does Not

IQ recovers most visibly after a move to family care, and earlier placement went with better outcomes in both cohorts, but recovery was partial. Brain volume, attention and social behavior differed in the ERA cohort even after the IQ impairment rate normalized. The page on whether you can improve your IQ places these results beside the interventions with the best evidence, and the page on how IQ changes with age describes what is stable across the lifespan.

6 Maltreatment and Violence in Ordinary Families: Cohorts, Twins and Baselines

Studies of abuse, neglect and violence in ordinary families find lower IQ in exposed children, but the estimate shrinks, and in two large birth cohorts nearly vanishes, once the analysis counts earlier ability and family background. The question is whether the exposure lowered the score or whether children who already scored lower, in families facing more adversity, were also more likely to be exposed.

Simple Associations

Perez and Widom compared 413 individuals with documented childhood abuse or neglect with 286 matched controls, tested for IQ at about age 28. The groups differed significantly after age, sex, race and social class were taken into account, and Danese and colleagues later summarized the comparison as a standardized difference of minus 0.62. Delaney-Black and colleagues assessed 299 urban first graders and, after controlling for sex, caregiver IQ, home environment, socioeconomic status and prenatal substance exposure, found violence exposure related to IQ. Their regression equation puts a child at or above the 90th percentile of both violence exposure and trauma related distress at 7.5 points, or 0.5 standard deviations, lower. Enlow and colleagues followed 206 children from low income families, recording interpersonal trauma from birth to 64 months and testing IQ at 24, 64 and 96 months. Exposure was associated with lower scores after adjustment for socioeconomic factors, maternal IQ, birth complications, birth weight and home stimulation, and children exposed in the first 2 years scored about one half standard deviation lower.

Twins and an Early Baseline

Koenen and colleagues analyzed 1,116 twin pairs aged 5 in England. Children exposed to high levels of maternally reported domestic violence had IQs about 8 points lower, in a dose response pattern, and the violence accounted for 4 percent of IQ variation independent of latent genetic influences.

Danese and colleagues added an IQ score from before most of the exposure. They used the E-Risk study of 2,232 British children followed to age 18 and the Dunedin study of 1,037 New Zealanders followed to age 38, with victimization recorded prospectively. In E-Risk, poly-victimization between ages 5 and 12 predicted lower IQ at 12, with a standardized coefficient of minus 0.17, which fell to minus 0.05 after adjustment for IQ at 5 and family status. At age 18 it fell from minus 0.12 to 0.00. In Dunedin, maltreatment between ages 3 and 11 predicted lower IQ at 11 to 13, with a coefficient of minus 0.11 that fell to 0.00 after adjustment for the mother's IQ, the child's IQ at 3 and family status. At age 38 it fell from minus 0.14 to minus 0.04, which was not significant. Before adjustment the standardized differences were 0.68 and 0.52 in E-Risk and 0.29 and 0.43 in Dunedin, close to the 0.62 of Perez and Widom. Within twin pairs, differences in victimization were not associated with differences in cognition, except for IQ at 12 in fraternal twins. The authors conclude that the association is largely noncausal and that lower cognitive ability may be a risk factor for victimization.

Why the Answers Differ

Koenen and Danese have overlapping authors and describe English twin samples of the same size, 1,116 pairs or 2,232 children, so they very likely draw on the same cohort, although the Koenen abstract does not name it. Koenen measured domestic violence before age 5 and IQ at 5. Danese measured victimization from 5 to 12 with IQ at 5 as the baseline. Danese and colleagues note that their results may not apply to victimization of infants and toddlers and that a small causal effect on the age 12 score in one sample cannot be ruled out. Enlow and colleagues adjusted for maternal IQ and still found about half a standard deviation after very early exposure, in a smaller low income sample. A modest effect of very early exposure, a larger apparent effect from unmeasured family factors and a near zero effect of later exposure could all be true at once, although none of the studies tests that directly. The page on what heritability means explains what twin, adoption and genomic evidence support.

In E-Risk, young adults who reported maltreatment on a questionnaire at 18 performed more poorly on executive function and processing speed tests but not on IQ tests, and those differences also attenuated. A meta-analysis by Matte-Landry and colleagues pooled 10 longitudinal studies and found IQ lower by 0.71 standard deviations in children with complex trauma, but only 2 of the 13 studies reviewed measured cognition before and after the trauma, so the figure cannot separate cause from preexisting difference.

7 Stress Before Birth and Chronic Stress in Adolescents and Adults

Studies of stress before birth and of chronic stress later in life report small associations with cognition, and whether those associations survive adjustment varies from study to study. Prenatal studies usually measure IQ, while adolescent and adult studies often measure one ability, so the numbers do not transfer.

Before Birth

Laplante and colleagues studied 89 children whose mothers were pregnant during the January 1998 ice storm in Quebec, which cut power for 3 million people for as long as 40 days. That June the mothers completed questionnaires on objective stress and subjective distress, and the children were tested at age 5 and a half with the Wechsler Preschool and Primary Scale of Intelligence and a picture vocabulary test. Children exposed in the womb to high levels of objective stress had lower Full Scale IQ, Verbal IQ and language scores than those exposed to low or moderate levels. Neither objective nor subjective stress was related to Performance IQ. A regional disaster differs from ordinary stress, but exposure within the storm was not randomly assigned, and the abstract does not give the size of the difference in points.

Schuurmans and colleagues used the population based Generation R cohort. In 2,401 children they estimated latent factors of prenatal stress and of postnatal stress from birth to age 10, and measured IQ with four subtests of the Wechsler Intelligence Scale for Children, fifth edition. Standardized coefficients were minus 0.20 for prenatal and minus 0.22 for postnatal stress, adjusted for sex and age. They are prospective associations, not experimental effects. The page on IQ and academic achievement describes how closely IQ and school results track each other.

Adolescents and Adults

Evans and Schamberg followed 195 participants whose mean age at testing was 17.3. Childhood poverty was the proportion of months from birth to age 13 at or below the poverty line, and chronic stress was allostatic load, a sum of six physiological markers measured at ages 9 and 13. Working memory was the longest sequence of items recalled in order on a touch pad task. Participants raised in middle income families averaged 9.44 items and those raised in poverty 8.50. Adding allostatic load cut the poverty coefficient from minus 1.01 to minus 0.77, a reduction of about a quarter (our arithmetic) that the authors report as statistically significant. The authors write that the data are not from an experiment.

Seitz and Steger analyzed 10,416 German adults aged 29 to 71. Chronic stress came from an 11 item questionnaire, and ability from a matrix reasoning test and a shortened picture vocabulary test. Small negative associations persisted after controlling for age, gender, income, retirement, household language and marital status, with latent correlations of minus 0.07 and minus 0.08 in the sensitivity analysis. The data are cross-sectional, so the direction is unknown. Elovainio and colleagues used the Whitehall II study of 4,146 British civil servants aged 35 to 55. Longer exposure to high strain jobs, defined as high demands with low control, went with lower reasoning, verbal and memory scores on tests given in 1997 to 1999 and 2003 to 2004, but those associations disappeared after adjustment for employment grade.

The pattern echoes the birth cohorts: raw associations that shrink or vanish once social position is counted, with the direction rarely settled by the design.

8 Acute Stress and the Day of the Test

Acute stress reliably impairs working memory in the laboratory, yet the one experiment reviewed here that followed induced stress with a full intelligence test found no change in total IQ. This is the state side of the question. Its evidence comes from short laboratory experiments and from one natural experiment that exploits the timing of local violence.

What Laboratory Stress Does to Executive Functions

Shields, Sazma and Yonelinas meta-analyzed 51 experiments in which participants without a known psychological or psychiatric disorder faced a validated laboratory stressor and then completed executive function tasks. Across 34 working memory experiments with 1,353 participants, stress impaired working memory by a standardized difference of 0.197, with a 95 percent interval from 0.064 to 0.330. Cognitive flexibility, in 6 experiments with 280 participants, was impaired by 0.300, and overall inhibition was not significantly affected. When working memory load was high the effect was 0.303, and when it was not high the effect was 0.049 and not significant. On a scale with a standard deviation of 15, 0.197 is about 3 points. That is our arithmetic and only an illustration, because these were laboratory working memory tasks and not IQ subtests. The page on working memory tests describes the construct, and Digit Span is one of its formats.

The Experiment That Used a Full Intelligence Test

Akan, Yildirim and Wolf published the experiment in 2025. Forty men aged 18 to 34 attended two sessions 28 days apart, in counterbalanced order. In one session they underwent the socially evaluated cold pressor test, with a hand in ice water while observed and filmed. In the other they handled warm water without observation. About 25 minutes later they began a German intelligence structure test of about 77 minutes, in parallel forms matched for difficulty. Cortisol and subjective ratings confirmed that the stress induction worked.

Stress had no significant effect on total IQ, and neither did trait test anxiety or the two together. The Bayes factor for the stress effect was 0.015, which the authors describe as very strong evidence for no effect. Scores were higher at the second session regardless of condition, a practice effect. The authors list their limits. The sample was small and could not detect small effects, it included only men, and the cold pressor test is a weaker stressor than the Trier Social Stress Test.

A Natural Experiment on Timing

Sharkey used children aged 5 to 17 in the Project on Human Development in Chicago Neighborhoods, comparing children in the same neighborhood who were assessed at different times relative to local homicides. Among African American children, the strongest results show that a homicide in the block group less than a week before the assessment reduced vocabulary and reading performance by about 0.5 and 0.66 standard deviations. The main results were replicated in a second Chicago data set. These were vocabulary and reading tests, not a full IQ battery, and the event was extreme. The 2013 scarcity study by Mani and colleagues in Science followed the same sugarcane farmers before and after harvest, so it is also a within person design, but it is not evidence about stress: the authors report that stress measures did not account for the effect, as the section on viral numbers explains. The page on what a low IQ means documents the study, and the page on poverty and IQ treats it as a question about money.

Putting the State Evidence Together

The three state findings, a modest working memory effect, a large vocabulary and reading effect after nearby violence and no change in total IQ, are compatible if state effects are specific to some abilities, to heavy loads, to strong stressors and to the first hours or days, or if a composite score is simply less sensitive than a single task. None of the three tells an individual whether a given score was affected. The page on sleep and IQ reaches the same kind of conclusion about a bad night.

9 How Stress Might Act: Mechanisms and What Mediation Does Not Prove

Proposed mechanisms run from stress hormones to caregiving quality to brain volume, and the evidence for each is mostly statistical mediation, not a demonstrated causal chain in people. A mediator statistically carries part of an association from exposure to outcome, which does not show that changing it would change the outcome, especially when it was measured and not assigned.

Lupien, McEwen, Gunnar and Heim argue in a review that chronic exposure to stress hormones, at any life stage from the prenatal period to old age, affects brain structures involved in cognition and mental health. They add that the effects depend on the timing and duration of exposure and on the interaction between genes and earlier adversity. Danese and colleagues note a tension. Animal experiments show causal effects of early life stress on the brain, but their human cohorts found deficits largely explained by earlier ability and family background. They suggest differences in life history across species, greater genetic variety in humans, buffering by schooling and child protection, and selective reporting of positive results.

The human mediation evidence is thinner. In Bucharest, caregiving quality and attachment security were the mediators. Total brain volume, 8.6 percent smaller, statistically mediated the institutional link to lower IQ in the ERA adults, and allostatic load accounted for about a quarter of a poverty association with working memory. None of these mediators was randomized, and none of the studies could exclude alternatives.

A different line of work asks what kind of adversity matters. Johnson and colleagues pooled 91 articles covering 82 cohorts and 31,188 individuals. Both deprivation and threat were associated with reduced executive functioning. The association with working memory was stronger for deprivation, 0.54 against 0.28, as was inhibitory control, 0.43 against 0.27, with no significant difference for cognitive flexibility. In CHC terms working memory is the broad ability Gwm, and the page on IQ and memory explains how it relates to a full score. These associations are observational and concern executive functions, not IQ. The Bucharest trial shows that placement in foster care changes IQ, with caregiving a candidate mediator that was not randomized, and the cohorts show that much of the association between adversity and IQ can be accounted for by what came before. The two conclusions concern different exposures, and neither identifies a biological pathway that an individual reader could check.

10 Points, Standard Deviations and Where the Viral Numbers Come From

Points and standard deviations convert only under assumptions, and each number that circulates about stress and IQ belongs to a specific design that the headline leaves out. On a scale with a standard deviation of 15, as the page on why the IQ scale uses a standard deviation of 15 explains, 0.5 standard deviations is 7.5 points. Many studies standardize within their own sample, where the standard deviation may differ from 15, so every multiplication on this page is our illustration and not a measurement. Conversions can also disagree inside one source. The abstract of the age 18 Bucharest paper gives its 9 point difference as 0.6 standard deviations, while its Table 1 lists 0.45 for the same contrast, and the orphanage meta-analysis gives a 20 point gap beside a trimmed standardized difference of 0.74, about 11 points at a standard deviation of 15. The 0.5 and 0.66 of the Chicago homicide study would be about 7.5 and 9.9 points, but of vocabulary and reading, not of IQ.

Each figure that circulates belongs to one exposure, one measure and one comparison.

• 13 points. This is printed in the 2013 Science scarcity study by Mani, Mullainathan, Shafir and Zhao, which says that, with a mean of 100 and a standard deviation of 15, the effects it observed correspond to about 13 IQ points, and that the laboratory effect matched the loss of a night of sleep while the field effect was three quarters of that. It is the authors' own approximation, not a measured change on a normed full battery, and the field result is a within person shift of about one item on a short matrices task (the page on what a low IQ means gives the figures). The same paper reports that, in a 2009 sample, farmers did show more stress before harvest on two biomarkers, heart rate and blood pressure, but that the harvest effect on cognition stayed significant and did not shrink when those measures were controlled, so the authors conclude that stress cannot fully explain it. A headline that says stress cost 13 points goes beyond the paper.
• 20 points. This is the orphanage meta-analysis, an average over 75 studies, which describes a gap between children's homes and families and not any child, and is not an estimate of what stress does.
• 15 points. This appears in the abstract of the age 11 follow-up of Romanian adoptees, for children over 6 months old on arrival from an institution.
• 8 points and 7.5 points. These are the twin study of domestic violence and the study of urban first graders. A news release distributed through PR Newswire, read on October 6, 2026, described the half standard deviation in the Minnesota cohort as the equivalent of 7 IQ points.

This review did not find a study that measured the points cost of ordinary stress on a full intelligence battery. The closest experiment, by Akan and colleagues in 2025, found no effect on total IQ after a laboratory stressor, and the distance between near zero and 20 points is a difference in what was studied.

11 What This Does Not Say About You, Your Child or a Bad Season

These studies describe groups and not individuals, and they cannot tell any reader how many IQ points a hard period cost. A difference of half a standard deviation between an exposed group and an unexposed group still leaves the two distributions heavily overlapping, so a group average is a poor guide to any one score. The page on what an IQ score means explains how to read a score against its reference group and its confidence interval.

The cohort evidence adds a limit for adults who look back. In the birth cohorts with an early baseline, children later victimized already scored lower before the exposure, so a lower score in an adult with a difficult history does not show that the history lowered it. The page on IQ and mental health covers how conditions and scores relate, and this page makes no diagnostic reading of anyone. The randomized and adoption evidence concerns institutions, far from most readers' childhoods, and the adult chronic stress studies find small associations with single abilities. None of it shows that a stressful decade lowered a particular person's IQ, and none of it shows that stress has no effect.

Taking a Test During a Hard Period

A person who is acutely stressed on the day may score lower on the most working memory dependent tasks, though this review found no study that quantifies that for adults taking an online battery. Retesting is not a neutral check, because scores rose at the second session in the 2025 experiment regardless of stress. The page on preparing for an IQ test covers practice effects.

A Disclosure About ACIS

We sell a paid, self-administered online assessment, so read this section as a disclosure. ACIS is unsupervised and online, is not a clinical or diagnostic instrument, is not suitable for hiring, school accommodations or admission to high IQ societies, and is offered in English only. It reports a Full Scale IQ and six index scores, including a working memory index built from Digit Span, Alphanumeric Sequencing and Visual Sequence, with percentiles and a 95 percent confidence interval. Breaks are allowed and a form can be completed over 30 days, so a person can choose when to sit each part, though no evidence reviewed here shows that this changes scores.

The working memory index is the one the laboratory literature would make most plausibly sensitive to a state effect, but those studies used other tasks, so their sizes cannot be transferred to ACIS subtests. The report cannot tell you whether a particular day lowered or raised your result, and it does not measure stress or adversity. The technical manual is the place for documentation, and the page on whether online IQ tests are accurate sets out what to trust and what to doubt in an online test.

12 What the Evidence Supports, Stated Narrowly

The evidence supports five narrow statements, and each one belongs to the design that produced it.

1. Severe, prolonged institutional deprivation is followed by substantially lower measured IQ. Randomized foster care raised IQ by about 9 points at age 18 in Bucharest without closing the gap to community peers, which was about 26 points in a comparison that was not randomized.
2. Observational adoption studies and meta-analyses find lower IQ on average after institutional rearing, a mean of 84 against 104, and better outcomes with earlier family care. IQ recovers most visibly, while brain volume, attention and social functioning differed lastingly in the English and Romanian adoptee cohort.
3. For maltreatment and violence in ordinary families, unadjusted standardized differences run from about 0.3 to 0.7, but in two birth cohorts with an early IQ baseline the adjusted associations fell to between minus 0.05 and 0.00. A smaller low income cohort still found about one half standard deviation, so a small causal effect cannot be excluded.
4. Acute laboratory stress impairs working memory by about 0.2 standard deviations, more under high load, and a homicide in the week before testing lowered vocabulary and reading scores in Chicago, but the one experiment with a full intelligence test found no effect on total IQ.
5. Chronic stress in adolescents and adults is associated with small differences in single abilities, confounded by social position or of unknown direction.

The evidence does not support converting any kind of stress into IQ points for an individual, dividing one person's score into a state part and a trait part, or claiming that ordinary stress permanently lowers IQ, and it does not support the opposite claim that stress never matters. Where the exposure is extreme, as in institutional rearing, the gaps are large and part of them can be closed. Where it is ordinary, the associations are small, uncertain and often explained by what came before.

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 definition, reading one score as a measure of how much stress lowered someone's ability is an interpretation the evidence above does not support. The American Psychological Association's Ethical Principles of Psychologists and Code of Conduct, in the version effective January 1, 2017, addresses the same point in Standard 9.06. It requires psychologists who interpret assessment results to take into account 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 to indicate any significant limitations. Citing these documents does not imply that their authors endorse ACIS or any other product.

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. Standardized differences, correlations, sample sizes and intervals are quoted as the papers report them, and any conversion into IQ points or any subtraction of published means is labeled as our arithmetic. For papers behind publisher paywalls, such as the Science, Lancet, Merrill-Palmer Quarterly and Psychological Bulletin papers, the figures come from the published abstracts, and the full text was consulted for the open access papers. The 13 point calibration and the stress finding were read in the text of the Mani paper, in a copy of the published article. All pages were read on October 6, 2026.

• Nelson C A, Zeanah C H, Fox N A, Marshall P J, Smyke A T and Guthrie D. Cognitive recovery in socially deprived young children: The Bucharest Early Intervention Project. Science, 2007, volume 318, issue 5858, pages 1937 to 1940.
• Fox N A, Almas A N, Degnan K A, Nelson C A and Zeanah C H. The effects of severe psychosocial deprivation and foster care intervention on cognitive development at 8 years of age: Findings from the Bucharest Early Intervention Project. Journal of Child Psychology and Psychiatry, 2011, volume 52, issue 9, pages 919 to 928; and Almas A N, Degnan K A, Nelson C A, Zeanah C H and Fox N A. IQ at age 12 following a history of institutional care: Findings from the Bucharest Early Intervention Project. Developmental Psychology, 2016, volume 52, issue 11, pages 1858 to 1866.
• Humphreys K L, King L S, Guyon-Harris K L, Sheridan M A, McLaughlin K A, Radulescu A, Nelson C A, Fox N A and Zeanah C H. Foster care leads to sustained cognitive gains following severe early deprivation. Proceedings of the National Academy of Sciences, 2022, volume 119, issue 38, article e2119318119; and King L S, Guyon-Harris K L, Valadez E A, Radulescu A, Fox N A, Nelson C A, Zeanah C H and Humphreys K L. A comprehensive multilevel analysis of the Bucharest Early Intervention Project: Causal effects on recovery from early severe deprivation. American Journal of Psychiatry, 2023, volume 180, issue 8, pages 573 to 583.
• Rutter M and the English and Romanian Adoptees Study Team. Developmental catch-up, and deficit, following adoption after severe global early privation. Journal of Child Psychology and Psychiatry, 1998, volume 39, issue 4, pages 465 to 476; Beckett C, Maughan B, Rutter M, Castle J, Colvert E, Groothues C, Kreppner J, Stevens S, O'Connor T G and Sonuga-Barke E J S. Do the effects of early severe deprivation on cognition persist into early adolescence? Findings from the English and Romanian Adoptees study. Child Development, 2006, volume 77, issue 3, pages 696 to 711; Sonuga-Barke E J S, Kennedy M, Kumsta R, Knights N, Golm D, Rutter M, Maughan B, Schlotz W and Kreppner J. Child-to-adult neurodevelopmental and mental health trajectories after early life deprivation: the young adult follow-up of the longitudinal English and Romanian Adoptees study. The Lancet, 2017, volume 389, issue 10078, pages 1539 to 1548; and Mackes N K, Golm D, Sarkar S, Kumsta R, Rutter M, Fairchild G, Mehta M A, Sonuga-Barke E J S and colleagues. Early childhood deprivation is associated with alterations in adult brain structure despite subsequent environmental enrichment. Proceedings of the National Academy of Sciences, 2020, volume 117, issue 1, pages 641 to 649.
• van IJzendoorn M H, Luijk M P C M and Juffer F. IQ of children growing up in children's homes: A meta-analysis on IQ delays in orphanages. Merrill-Palmer Quarterly, 2008, volume 54, issue 3, pages 341 to 366; and van IJzendoorn M H, Juffer F and Poelhuis C W K. Adoption and cognitive development: A meta-analytic comparison of adopted and nonadopted children's IQ and school performance. Psychological Bulletin, 2005, volume 131, issue 2, pages 301 to 316.
• Danese A, Moffitt T E, Arseneault L, Bleiberg B A, Dinardo P B, Gandelman S B, Houts R, Ambler A, Fisher H L, Poulton R and Caspi A. The origins of cognitive deficits in victimized children: Implications for neuroscientists and clinicians. American Journal of Psychiatry, 2017, volume 174, issue 4, pages 349 to 361; and Koenen K C, Moffitt T E, Caspi A, Taylor A and Purcell S. Domestic violence is associated with environmental suppression of IQ in young children. Development and Psychopathology, 2003, volume 15, issue 2, pages 297 to 311.
• Perez C M and Widom C S. Childhood victimization and long-term intellectual and academic outcomes. Child Abuse and Neglect, 1994, volume 18, issue 8, pages 617 to 633; Delaney-Black V, Covington C, Ondersma S J, Nordstrom-Klee B, Templin T, Ager J, Janisse J and Sokol R J. Violence exposure, trauma, and IQ and/or reading deficits among urban children. Archives of Pediatrics and Adolescent Medicine, 2002, volume 156, issue 3, pages 280 to 285; Bosquet Enlow M, Egeland B, Blood E A, Wright R O and Wright R J. Interpersonal trauma exposure and cognitive development in children to age 8 years: a longitudinal study. Journal of Epidemiology and Community Health, 2012, volume 66, issue 11, pages 1005 to 1010; and the news release Maltreatment or Witnessing Family Violence Can Lower a Child's IQ, PR Newswire.
• Matte-Landry A, Grisé Bolduc M-È, Tanguay-Garneau L, Collin-Vézina D and Ouellet-Morin I. Cognitive outcomes of children with complex trauma: A systematic review and meta-analyses of longitudinal studies. Trauma, Violence, and Abuse, 2023, volume 24, issue 4, pages 2743 to 2757; and Johnson D, Policelli J, Li M, Dharamsi A, Hu Q, Sheridan M A, McLaughlin K A and Wade M. Associations of early-life threat and deprivation with executive functioning in childhood and adolescence: A systematic review and meta-analysis. JAMA Pediatrics, 2021, volume 175, issue 11, article e212511.
• Laplante D P, Brunet A, Schmitz N, Ciampi A and King S. Project Ice Storm: Prenatal maternal stress affects cognitive and linguistic functioning in 5½-year-old children. Journal of the American Academy of Child and Adolescent Psychiatry, 2008, volume 47, issue 9, pages 1063 to 1072; and Schuurmans I K, Luik A I, de Maat D A, Hillegers M H J, Ikram M A and Cecil C A M. The association of early life stress with IQ-achievement discrepancy in children: A population-based study. Child Development, 2022, volume 93, issue 6, pages 1837 to 1847.
• Evans G W and Schamberg M A. Childhood poverty, chronic stress, and adult working memory. Proceedings of the National Academy of Sciences, 2009, volume 106, issue 16, pages 6545 to 6549; Seitz M and Steger D. The mind under pressure: What roles does education play in the relationship between chronic stress and cognitive ability? Journal of Intelligence, 2025, volume 13, issue 2, article 13; and Elovainio M, Ferrie J E, Singh-Manoux A, Gimeno D, De Vogli R, Shipley M J, Vahtera J, Brunner E J and colleagues. Cumulative exposure to high-strain and active jobs as predictors of cognitive function: the Whitehall II study. Occupational and Environmental Medicine, 2009, volume 66, issue 1, pages 32 to 37.
• Shields G S, Sazma M A and Yonelinas A P. The effects of acute stress on core executive functions: A meta-analysis and comparison with cortisol. Neuroscience and Biobehavioral Reviews, 2016, volume 68, pages 651 to 668; Akan O, Yildirim M and Wolf O T. The reassuring absence of acute stress effects on IQ test performance. Journal of Intelligence, 2025, volume 13, issue 10, article 131; and Sharkey P. The acute effect of local homicides on children's cognitive performance. Proceedings of the National Academy of Sciences, 2010, volume 107, issue 26, pages 11733 to 11738.
• Lupien S J, McEwen B S, Gunnar M R and Heim C. Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature Reviews Neuroscience, 2009, volume 10, issue 6, pages 434 to 445; McLaughlin K A, Sheridan M A and Lambert H K. Childhood adversity and neural development: Deprivation and threat as distinct dimensions of early experience. Neuroscience and Biobehavioral Reviews, 2014, volume 47, pages 578 to 591; Shonkoff J P, Garner A S and the committees named on the report. The lifelong effects of early childhood adversity and toxic stress. Pediatrics, 2012, volume 129, issue 1, pages e232 to e246; and the Center on the Developing Child at Harvard University, What is toxic stress?, page read October 6, 2026.
• Mani A, Mullainathan S, Shafir E and Zhao J. Poverty impedes cognitive function. Science, 2013, volume 341, issue 6149, pages 976 to 980.
• 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