Cannabis and IQ Decline in Youth

1,738 words · 7 min read · 4 references cited

Heavy or dependent cannabis use during adolescence is associated with a small but measurable decline in IQ — about 2 points on average, concentrated in verbal ability. The effect is real, but it is not the 8-point loss often quoted from a single 2012 study. Twin studies and confounder-adjusted cohort analyses suggest part of the association is explained by family background, co-use of other substances, and pre-existing differences rather than cannabis itself. The honest answer to “does cannabis lower IQ?” is: probably, modestly, in the heaviest adolescent users — and the size of the effect depends heavily on how the question is framed.

What “lowers IQ” means in this research

IQ studies of cannabis use almost always rely on a longitudinal design: investigators measure cognitive ability at a baseline (typically before regular cannabis use begins), then again years later. The “decline” reported is the within-person change, often compared to a non-using control group. Because adolescence and early adulthood are themselves periods of cognitive growth, a flat or modestly declining trajectory in a user is interpreted against the rising trajectory of a non-user. Effect sizes are usually expressed as Cohen’s d or as IQ-point equivalents on the standard 100-point, 15 SD scale.

This framing matters. A 2-point average decline does not mean every cannabis user loses 2 points; it means the user distribution sits 2 points lower than the comparison distribution after follow-up.

What the longitudinal evidence shows

Three lines of evidence dominate this question, and they do not all point the same way.

Dunedin, heavy adolescent-onset8 pointsPower meta-analysis (pooled)~2 pointsALSPAC, confounder-adjusted0.1 points (n.s.)02468Reported IQ decline (points)
Figure 1. The cannabis–IQ effect shrinks as studies control confounders more tightly: from 8 points in one small Dunedin subgroup to about 2 in meta-analysis and 0.1 once fully adjusted (Meier 2012; Power 2021; Mokrysz 2016).

The Dunedin cohort (Meier et al., 2012)

The most-cited finding came from the Dunedin Multidisciplinary Health and Development Study, a New Zealand cohort followed from age 13 to age 38. Heavy users who began before age 18 and continued into adulthood lost an average of 8 IQ points by midlife. The result drove a wave of public health messaging — but the subgroup was small: only 19 participants met the criterion of clinical dependence at multiple assessments. Quitting did not reverse the decline.

The Power meta-analysis (2021)

A more comprehensive picture comes from Power et al.’s systematic review and meta-analysis, which pooled seven longitudinal cohort studies with 808 frequent or dependent cannabis users and 5,308 controls. The pooled effect was a Cohen’s d of −0.132 (95% CI −0.198 to −0.066), translating to roughly a 2-point IQ decline. Heterogeneity across studies was low (I² = 0.2%), meaning the seven studies agreed with each other unusually well.

A subgroup analysis revealed something the simple headline conceals: the decline is concentrated in verbal IQ (d = −0.196, ~3-point decline), while performance IQ showed no significant change (d = −0.004). Whatever cannabis is doing to adolescent cognition, it is not affecting all cognitive domains equally.

The confounder-adjusted cohorts and twin studies

Two studies cast the most doubt on a direct causal interpretation. Mokrysz and colleagues, analyzing the UK ALSPAC cohort (n = 2,235), found that adolescents with at least 50 lifetime cannabis uses by age 15 had IQ scores only 0.1 points lower than never-users — a non-significant difference — once cigarette and alcohol use, childhood mental-health symptoms, and behavioural problems were entered as covariates. Jackson et al. (2016), comparing marijuana-using adolescents to their abstinent twin siblings across two longitudinal twin samples, found that users showed no greater IQ decline than their non-using siblings — a result that points to shared family-level factors as the more parsimonious explanation.

The takeaway: associations between cannabis and IQ are robust, but the more rigorously confounders are controlled, the smaller and less specific the cannabis-attributable effect becomes.

Concentration of the decline in verbal IQ

The Power meta-analysis’s verbal/performance split is one of the most informative findings in this literature. Verbal IQ taxes crystallized abilities — vocabulary, comprehension, general knowledge — that depend on accumulated learning. Performance IQ leans on fluid abilities, processing speed, and visual-spatial reasoning. If chronic cannabis exposure during adolescence reduced engagement with academic content (missed classes, lower study time, reduced reading), the resulting deficit would appear as slower acquisition of verbal knowledge rather than damage to general processing capacity. This is consistent with what is observed.

Verbal IQ (crystallized)d = −0.196 (~3 pts)Performance IQ (fluid)d = −0.004 (n.s.)-0.2-0.15-0.1-0.050Cannabis effect on IQ (Cohen's d)
Figure 2. The decline concentrates in verbal (crystallized) ability rather than performance (fluid) ability — a pattern consistent with educational disengagement (Power et al., 2021).

In other words, the IQ decline associated with adolescent cannabis use may reflect educational disengagement as much as direct neurotoxicity.

The confounding problem, and why it matters

Several factors make this question difficult to answer with observational data:

  • Selection effects. The adolescents who initiate heavy cannabis use differ systematically from those who do not, in temperament, family environment, peer group, mental health, and academic engagement.
  • Co-use. Heavy adolescent cannabis users are also far more likely to use alcohol, nicotine, and other illicit substances. When investigators control for binge drinking and polysubstance use, cannabis-specific cognitive effects often shrink toward zero.
  • Reverse causation. Lower-than-average childhood IQ is itself a risk factor for early substance use, meaning some of the cross-sectional gap predates cannabis exposure.
  • Family-level confounders. This is what twin studies isolate. The Jackson et al. (2016) finding — abstinent twins of users show similar IQ patterns — is hard to reconcile with a strong direct causal effect.

The most defensible reading of the literature is that frequent or dependent cannabis use in adolescence has a small direct effect on verbal IQ, partially confounded with broader risk factors that cluster with heavy use.

Plausible biological mechanisms

A direct neurobiological pathway exists. The endocannabinoid system, particularly CB1 receptors, is densely expressed in the hippocampus and prefrontal cortex — regions undergoing extensive synaptic pruning and myelination during adolescence. THC’s interaction with these receptors during a critical developmental window could plausibly disrupt circuits supporting memory consolidation and executive function. Animal studies support this mechanism, and human imaging studies report reduced hippocampal volume and altered prefrontal connectivity in heavy adolescent users. The mechanism is biologically reasonable; it does not, by itself, prove the size or specificity of the effect at the population level.

What this means for parents, educators, and young users

Translating this evidence into guidance:

  • Frequency and dependence matter more than any use. The studies that find an effect almost always identify heavy or clinically dependent users as the affected group. Casual or experimental use shows much weaker associations.
  • Age of onset matters. Beginning regular use before 18 is associated with worse cognitive outcomes than beginning later. The adolescent brain’s ongoing development is the proposed reason.
  • Two IQ points is small at the individual level, meaningful at the population level. A 2-point shift will not be detectable in any single person, but it represents a real shift in the distribution.
  • The rest of the risk profile matters. Adolescents using cannabis heavily are typically also experiencing other risk factors. Addressing those holistically is more useful than focusing on cannabis alone.

What we still do not know

Several questions remain unresolved:

  • Reversibility. Most longitudinal studies have only mid-adulthood follow-up. Whether IQ recovers fully after extended abstinence is not well established.
  • High-THC products. The cannabis used in cohort studies whose participants started in the 1990s and 2000s was substantially weaker than current legal-market flower and concentrates. Effects of modern products on adolescent brains are largely unmeasured.
  • Edibles versus inhalation. Most evidence comes from smoked cannabis. Whether absorption route changes long-term cognitive impact is unknown.
  • Below-the-threshold effects. Subclinical effects on attention, memory, or executive function — not captured by general IQ tests — could exist even where IQ measures show nothing.

Frequently Asked Questions

Does cannabis permanently lower IQ?

The Dunedin cohort found that quitting did not restore lost IQ points in heavy adolescent-onset users. Other studies are less consistent. The honest answer is that some persistent effect appears likely in heavy users, but recovery dynamics are not well-characterized.

At what age is cannabis most harmful to cognition?

Use beginning before 18 — and especially before 16 — is associated with worse outcomes than later-onset use. The proposed reason is that adolescence is a period of active synaptic pruning and myelination, particularly in prefrontal regions involved in executive function.

How much cannabis counts as “heavy” or “dependent”?

The studies finding effects typically classify “heavy” as daily or near-daily use over months to years, and “dependent” using DSM criteria for cannabis use disorder (loss of control, tolerance, withdrawal, continued use despite consequences). Weekly or less-frequent use rarely produces measurable IQ effects in the literature.

Is medical marijuana different?

The IQ research has mostly studied recreational use in adolescents. Medical use under physician supervision in adults is a different exposure profile and not directly addressed by these studies. The developmental concern specifically targets adolescent brains.

Can IQ recover after long-term abstinence?

Short-term cognitive effects (within hours and days of last use) resolve. Whether the IQ-level effects observed in heavy adolescent-onset users fully reverse with sustained abstinence remains an open question; the available follow-up data suggest partial but not complete recovery.

Does CBD affect IQ?

Cannabidiol (CBD) does not produce the psychoactive effects of THC and has not been shown to lower IQ in adolescents in the existing literature. The IQ research overwhelmingly concerns THC-containing cannabis products.

References

  • Power, E., Sabherwal, S., Healy, C., O’Neill, A., Cotter, D., & Cannon, M. (2021). Intelligence quotient decline following frequent or dependent cannabis use in youth: a systematic review and meta-analysis of longitudinal studies. Psychological Medicine, 51(2), 194–200. https://doi.org/10.1017/S0033291720005036
  • Meier, M. H., Caspi, A., Ambler, A., Harrington, H., Houts, R., Keefe, R. S. E., McDonald, K., Ward, A., Poulton, R., & Moffitt, T. E. (2012). Persistent cannabis users show neuropsychological decline from childhood to midlife. Proceedings of the National Academy of Sciences, 109(40), E2657–E2664. https://doi.org/10.1073/pnas.1206820109
  • Mokrysz, C., Landy, R., Gage, S. H., Munafò, M. R., Roiser, J. P., & Curran, H. V. (2016). Are IQ and educational outcomes in teenagers related to their cannabis use? A prospective cohort study. Journal of Psychopharmacology, 30(2), 159–168. https://doi.org/10.1177/0269881115622241
  • Jackson, N. J., Isen, J. D., Khoddam, R., Irons, D., Tuvblad, C., Iacono, W. G., McGue, M., Raine, A., & Baker, L. A. (2016). Impact of adolescent marijuana use on intelligence: Results from two longitudinal twin studies. Proceedings of the National Academy of Sciences, 113(5), E500–E508. https://doi.org/10.1073/pnas.1516648113

Related questions

What is executive function in children?

Executive function is the umbrella term for the cognitive control processes that allow children to manage their attention, hold information in mind, resist impulses, and shift flexibly between tasks. It is one of the strongest single predictors of how children do in school, in friendships, and decades later in adult life — stronger in some respects than IQ. The good news is that executive function is more malleable than general intelligence, and a well-replicated body of intervention evidence shows what kinds of activities actually move it. The catch is that "executive function training" is also one of the most over-marketed claims in child development, and most commercial brain-training products do not produce the outcomes they advertise. Read more →

How does screen time affect children's intelligence?

Few topics generate more parental anxiety than screen time, and few are reported with less precision. Headlines swing between "screens are rewiring children's brains" and "no evidence screens harm cognition" — both citing real studies, both correct in narrow ways, neither capturing what the research actually shows. The honest summary: screen-time effects on children's cognitive development are real but small, depend heavily on what is being viewed and what activities the screens displace, are strongest in the first three years of life, and are confounded with socioeconomic factors that account for some — but not all — of the observed associations. Read more →

How does sleep affect children's IQ?

Sleep is the most underrated lever in children's cognitive development. Parents pour money into tutoring, enrichment activities, and educational technology while leaving the most evidence-supported intervention sitting on the table — adequate, consistent, age-appropriate sleep. The largest contemporary study to address the question, Yang and colleagues' 2022 propensity-score-matched analysis of more than 8,300 children in the Adolescent Brain Cognitive Development (ABCD) cohort, found that pre-adolescents (age 9–10) sleeping less than nine hours per night showed measurable cognitive deficits compared with their adequately-rested matched peers — and that the deficit was about twice as large for crystallized intelligence as for fluid intelligence, was mediated by structural differences in specific brain regions, and persisted unchanged at two-year follow-up. The implications carry past parenting advice into a real public-health conversation about how much modern children actually sleep and what that costs them cognitively. Read more →

How does maternal milk affect neurodevelopment?

In neonatal intensive care units across the world, mothers of babies born too early are routinely encouraged to express breast milk to be fed to their infants by tube — sometimes for weeks or months before the infant can suckle directly. The clinical justification is partly nutritional, partly immunological, and partly cognitive: a substantial body of research suggests that babies who receive more of their mother's own milk during this critical period have better cognitive and academic outcomes years later. A 2022 study by Belfort and colleagues in JAMA Network Open provides the most recent dose-response evidence: among 586 Australian preterm infants followed to age 7, every additional 25 mL/kg/day of maternal milk during NICU hospitalization was associated with measurable improvements in performance IQ, reading, math, and ADHD symptom scores. The effect was strongest for the most premature infants and showed a striking domain-specificity: full-scale IQ, verbal IQ, executive function, and behavior were not affected. Read more →

What does "lowers IQ" actually mean in this research?

IQ studies of cannabis use almost always rely on a longitudinal design: investigators measure cognitive ability at a baseline (typically before regular cannabis use begins), then again years later. The "decline" reported is the within-person change, often compared to a non-using control group. Because adolescence and early adulthood are themselves periods of cognitive growth, a flat or modestly declining trajectory in a user is interpreted against the rising trajectory of a non-user. Effect sizes are usually expressed as Cohen's d or as IQ-point equivalents on the standard 100-point, 15 SD scale.

What does the longitudinal evidence actually show?

Three lines of evidence dominate this question, and they do not all point the same way. The most-cited finding came from the Dunedin Multidisciplinary Health and Development Study, a New Zealand cohort followed from age 13 to age 38. Heavy users who began before age 18 and continued into adulthood lost an average of 8 IQ points by midlife. The result drove a wave of public health messaging — but the subgroup was small: only 19 participants met the criterion of clinical dependence at multiple assessments. Quitting did not reverse the decline.