From Teen Years to 38: Heavy, Long-Term Cannabis Use Tracked With Lower Cognitive Scores

The Dunedin Study

Before the cannabis finding, you need to understand the dataset. The Dunedin Multidisciplinary Health and Development Study is one of the most extraordinary research projects in the history of psychology.

The cannabis-IQ paper was led by Madeline Meier, then a graduate student in Caspi and Moffitt's lab at Duke. It would become one of the most cited papers the Dunedin Study has ever produced.

What They Found

Cannabis use was assessed in confidential interviews at five time points: ages 18, 21, 26, 32, and 38. Participants were classified by their pattern of use — never used, used but not dependent, or cannabis-dependent at one or more assessment waves. IQ was measured at age 13 and again at age 38, using a comprehensive neuropsychological battery covering attention, memory, processing speed, verbal comprehension, and perceptual reasoning.

The decline was broad — not limited to one cognitive domain but spanning attention, memory, executive function, and processing speed. Friends and family of persistent users independently reported noticing more cognitive problems in daily life, providing external validation beyond test scores.

The finding that stopped everyone: cessation of cannabis use did not fully restore neuropsychological functioning among adolescent-onset users. Even those who had quit by age 38 still showed residual cognitive deficits compared to their own age-13 baseline.

But the critical nuance — often lost in media coverage — was the adolescent specificity. Adult-onset users showed no significant IQ decline. The vulnerability appeared to be specific to the developing brain.

The Firestorm

The paper was published online on August 27, 2012. Within months, three major critiques emerged.

The Rogeberg critique was particularly elegant. He didn't challenge the data — he built a mathematical model showing that if socioeconomic status affects both the likelihood of cannabis use and the trajectory of cognitive development (both plausible), the observed pattern of results could emerge without any causal effect of cannabis on IQ. The correlations were "consistent with confounding," he argued — not proof of confounding, but a demonstration that it couldn't be ruled out.

“It would be wrong to interpret these results as a direct causal finding. The observed correlations between persistent cannabis use and IQ decline are consistent with a model of confounding from socioeconomic status.”

The Moffitt team's response was pointed: in their data, the socioeconomic pathway Rogeberg proposed didn't actually exist. Low SES didn't predict adolescent cannabis dependence, and SES wasn't related to IQ decline. The confounding model was mathematically possible but empirically unsupported in the actual cohort.

The Twin Study Challenge

The most significant challenge came not from statisticians but from geneticists. In 2016, Nicholas Jackson and colleagues studied 3,066 twins — pairs where one twin used cannabis and the other did not. If cannabis causes cognitive decline, the using twin should perform worse than the non-using twin. They share the same genes, the same family environment, and (in the case of identical twins) the same DNA.

The twin study didn't disprove the Meier finding — but it offered a powerful alternative explanation. The teenagers who use cannabis heavily may already be on a different cognitive trajectory before they ever touch the drug. They may come from families where both cannabis exposure and cognitive development are influenced by the same genetic and environmental factors. If that's the case, cannabis isn't causing the decline — it's a marker for other risk factors.

Meier's team responded with their own co-twin analysis from the Dunedin data (2017), finding some cannabis-specific effects even after controlling for shared family factors. The debate remains genuinely unresolved.

Where the Field Landed

By 2021, enough longitudinal studies had accumulated for a definitive meta-analysis. Power and colleagues pooled seven studies encompassing over 6,000 participants and arrived at a revised estimate.

Two IQ points is not nothing. During active neurodevelopment, even small cognitive effects may compound over time. But it's a very different story from the 8-point headline. And the meta-analysis confirmed what Meier's original data showed: the effect was concentrated in youth, not adult users.

Why It Still Matters

The Meier study's lasting contribution isn't the specific number — it's the question it forced the field to take seriously. Before 2012, the idea that cannabis might have lasting cognitive effects was largely confined to animal research and small cross-sectional studies. After 2012, adolescent cannabis exposure became one of the most intensively studied questions in developmental neuroscience.

Myth vs. Reality

✕Myth

Cannabis causes an 8-point IQ decline — it's settled science.

✓Reality

The 8-point figure applies to the most extreme group in one study (persistent adolescent-onset dependent users in the Dunedin cohort). Meta-analyses across multiple studies estimate the average effect at roughly 2 IQ points. A powerful twin study found no within-pair differences, suggesting family factors may explain much of the association. Heavy adolescent use is associated with real cognitive costs — but the magnitude and the causal mechanism remain debated.

The Evidence

Meier et al. (2012): 8 points in most persistent users. Jackson et al. (2016): no within-pair twin differences. Power et al. (2021): pooled meta-analytic estimate of ~2 IQ points. All researchers agree: adult-onset use shows no significant IQ decline.

Meier et al. (2012), PNAS; Jackson et al. (2016), PNAS; Power et al. (2021), Psychol Med

The study also revealed something the anti-cannabis narrative often ignores: adult-onset users showed no significant cognitive decline. This finding is just as important as the adolescent finding, and it has direct implications for policy. Age restrictions — not prohibition — appear to be the evidence-based approach. The developing brain is vulnerable. The adult brain is far more resilient.

For parents navigating conversations about teen cannabis use, the honest message isn't "cannabis will definitely lower your IQ." It's: "the research shows real risks to the developing brain, the debate is about how big those risks are, and the safest course is to delay use until adulthood."

And for adults who used cannabis as teenagers and worry about permanent damage — the evidence on cognitive recovery is genuinely reassuring. The Scott et al. meta-analysis found that most cognitive deficits resolve with sustained abstinence. The brain is more plastic than the Meier headlines suggested.

The Researchers

Madeline Meier was a graduate student at Duke when this paper made her name one of the most recognized in cannabis research. She has since continued studying cannabis and cognition, testifying before the U.S. Senate Drug Caucus in 2019 — presenting a more nuanced picture than the original headlines allowed.

The Dunedin Study itself is led by Richie Poulton at the University of Otago, with Avshalom Caspi and Terrie Moffitt directing much of the analysis from Duke. It has been running for over 50 years. The 1,037 participants are now in their early 50s, still being assessed, still producing data that shapes our understanding of human development. This cannabis paper is one of hundreds the study has generated — but it remains one of the most consequential and most debated.

Frequently Asked Questions

Did this study prove cannabis causes IQ decline?

No. It showed an association between persistent adolescent-onset cannabis use and lower IQ scores at age 38 compared to age 13. As an observational study, it cannot prove causation. A subsequent twin study (Jackson et al., 2016) found no within-pair cognitive differences between cannabis-using and non-using twins, suggesting shared family and genetic factors may explain much of the association. The causal question remains genuinely debated among researchers.

How many IQ points did cannabis users actually lose?

The headline figure of 8 IQ points applied only to the most extreme group — participants who became cannabis-dependent in adolescence and remained dependent at three or more assessment waves over 20 years. Less persistent adolescent users showed smaller declines (~5 points). Adult-onset users showed no significant decline. A 2021 meta-analysis across seven studies estimated the average effect at roughly 2 IQ points — statistically significant but far smaller than the headline number.

Did adult-onset users show any cognitive decline?

No. This is one of the study's most important findings. Cannabis use beginning in adulthood — even persistent, heavy use — was not associated with significant IQ decline. The vulnerability appears specific to the adolescent brain during its final development period. This finding supports age-based policy approaches rather than blanket prohibition.

Does quitting cannabis restore IQ?

For adolescent-onset users in this study, cessation did not fully restore function by age 38. However, the Scott et al. (2018) meta-analysis of 69 studies found that most cognitive deficits resolve within 72 hours of cessation in adult users, with continued improvement over weeks. The picture for adolescent-onset users may be different, but the recovery data for adults is encouraging.

What did the twin study challenge show?

Jackson et al. (2016) studied 3,066 twins where one twin used cannabis and the other didn't. If cannabis directly damages the brain, the using twin should perform worse. They found no significant within-pair cognitive differences. This suggests that family background and genetic factors — rather than cannabis itself — may explain much of the cognitive association. The finding is the strongest challenge to the causal interpretation of the Meier results. However, the twin study used different cognitive measures, a different population, and a shorter follow-up than the Dunedin study.

Is this study still relevant with today's higher-potency cannabis?

The Dunedin cohort used cannabis typical of the 1980s-2000s — generally under 10% THC, with more CBD than modern products. Today's cannabis regularly exceeds 25% THC with near-zero CBD. If the Meier findings reflect genuine neurotoxicity, modern products could pose even greater risk. If they reflect confounding rather than direct effects, potency may be less relevant. Either way, the adolescent vulnerability window remains the most important finding.