Study Finds Long-Term Cannabis Use May Cause Physical Changes to Brain Structure

The March 2026 European Congress of Psychiatry in Prague opened with a provocative headline: long-term cannabis use is linked to measurable thinning of the frontal cortex [1]. Researchers from Spain's FIDMAG Germanes Hospitalàries Research Foundation reported that heavy, sustained cannabis users showed reduced cortical thickness in the right rostral middle frontal cortex — a region central to executive functioning, planning, and decision-making [1].

But this finding arrived in a research landscape that is anything but settled. Across at least five major studies published since 2024, the evidence on cannabis and brain structure ranges from alarming to reassuring, depending on which dataset you examine, which population was studied, and — critically — whether the analytical methods can distinguish correlation from causation.

The Studies: What Was Found, and in Whom

Three large-scale studies anchor the current debate.

The largest observational neuroimaging study to date, published in BMJ Mental Health in October 2024, drew on 15,896 participants from the UK Biobank — 3,641 lifetime cannabis users (mean age 61) and 12,255 controls (mean age 64.5) [2]. Using structural MRI, the researchers found that cannabis users had poorer white matter integrity in the corpus callosum, as well as weaker resting-state functional connectivity in default mode and central executive network regions [2]. Women showed changes across 24 brain structures, while men showed changes in 6 [2].

In January 2025, a study published in JAMA Network Open examined over 1,003 young adults aged 22 to 36, making it one of the largest functional imaging studies in this age group [3]. Lead researcher Joshua Gowin of the University of Colorado found that 63% of heavy lifetime cannabis users (those with more than 1,000 lifetime uses) showed reduced brain activity during working memory tasks, with 68% of recent users showing similar reductions [3]. The affected regions — dorsolateral prefrontal cortex, dorsomedial prefrontal cortex, and anterior insula — are involved in decision-making, memory, attention, and emotional processing [3]. Only working memory showed statistically significant effects across all seven cognitive tasks tested [3].

Then, in February 2026, a team led by Anika Guha at the University of Colorado Anschutz published findings from the single largest sample to date: 26,362 UK Biobank participants aged 40 to 77 [4]. Their results cut against the grain. Moderate cannabis use was associated with larger brain volumes across most regions studied, particularly in areas with high CB1 cannabinoid receptor density, including the hippocampus [4]. "Every cognitive measure that demonstrated a significant effect showed better performance among cannabis users," Guha reported. "It goes against your default assumptions" [4].

Source: Multiple peer-reviewed studies

Data as of Mar 1, 2026CSV

The variation in sample sizes across these studies matters. Smaller studies (under 200 participants) have historically produced more inconsistent results, with some finding reduced orbitofrontal cortex volume and others finding no differences [5]. The field's move toward biobank-scale datasets represents a methodological improvement, though larger samples also introduce their own complexities.

The Causation Problem

The most consequential finding in recent cannabis neuroimaging research may not be about the brain at all — it's about methodology.

The UK Biobank study published in BMJ Mental Health paired its observational analysis with Mendelian randomisation, a genetic technique that uses inherited variants as proxies for exposure to test whether an association is causal [2]. The result: "Genetic analyses found no support for causal relationships underlying these observed associations" [6]. The brain differences observed in cannabis users may be explained by unmeasured confounders — family history, diet, medications, socioeconomic factors — rather than cannabis itself [6].

This finding aligns with a broader pattern in the literature. A 2022 study in Neuroscience & Biobehavioral Reviews found that differences in structural brain integrity previously attributed to cannabis were better explained by polysubstance use, particularly concurrent alcohol and tobacco consumption [7]. When researchers controlled for other substance use, the cannabis-specific signal often weakened or disappeared.

The baseline rate of the structural variations seen in cannabis users also exists in the general population, complicating interpretation. No study to date has demonstrated that the observed brain differences were absent before cannabis use began. The UK Biobank study's authors acknowledged this limitation directly, noting that their predominantly healthy, White sample with few cannabis use disorders limits generalizability [6].

Adolescents vs. Adults: A Critical Distinction

If there is an emerging consensus in this literature, it centers on age. The evidence for adolescent vulnerability is more consistent than for adult use.

The Adolescent Brain Cognitive Development (ABCD) Study, funded by the National Institutes of Health and tracking over 11,000 youth aged 9 to 17, has produced several relevant findings [8]. Adolescents who initiated substance use had thinner cortical mantles in the prefrontal cortex compared to non-initiators, though they showed thicker cortices in temporal, occipital, and parietal regions [8]. Hair toxicology testing of 123 cannabis-positive youth (ages 13-14) showed lower scores on picture memory tests compared to matched controls [8].

Research from the University of Texas at Dallas found that participants who began using cannabis at age 16 or earlier showed signs of arrested prefrontal cortex development, while those who started after 16 showed the opposite pattern — signs of accelerated brain aging [9]. Animal models support this threshold: synthetic cannabinoids administered during early and mid-adolescence in rats interfered with GABA neurotransmitter function in the adult prefrontal cortex, but the same compounds given in late adolescence or adulthood did not produce this effect [9].

A Bayesian causal network modeling study published in Translational Psychiatry in 2022 suggested that adolescent cannabis use accelerates prefrontal cortical thinning, providing some of the stronger evidence for a directional relationship [10].

Source: OpenAlex

Data as of Jan 1, 2026CSV

The surge in academic publications on cannabis and brain structure — from 489 papers in 2011 to a peak of 4,274 in 2023 — reflects the urgency researchers feel about answering these questions before policy outpaces evidence [11].

Reversibility: What Happens After You Stop

The evidence on recovery is cautiously optimistic for adults and less clear for adolescents.

Cognitive deficits in attention, memory, and psychomotor speed often begin recovering within days of abstinence [12]. A meta-review of meta-analyses published in Addiction in 2022 found no significant effects on global cognition or any of eight neurocognitive domains after approximately one month of abstinence, suggesting no lasting residual effects for most adult users [13].

At the neurobiological level, the endocannabinoid system shows signs of recovery: downregulation of CB1 receptors can reverse after abstinence, restoring normal signaling sensitivity [12]. Structural changes, such as reduced hippocampal volume, have been shown to reverse after prolonged abstinence, though this process may take over two years [12].

The picture is less encouraging for adolescent-onset users. The landmark Dunedin longitudinal study found that persistent cannabis use starting in adolescence was associated with IQ declines that did not fully recover after cessation [14]. However, the clinical significance of these findings remains debated: a 2018 meta-analysis of 69 studies in JAMA Psychiatry found that associations between cannabis use and cognitive functioning in adolescents and young adults were "small and may be of questionable clinical importance for most individuals" [15].

The Steelman Case for Neutral or Positive Effects

The Guha et al. (2026) findings — larger brain volumes and better cognitive performance in cannabis users aged 40-77 — represent the strongest data point for those arguing that structural brain changes associated with cannabis may be neurologically neutral or even beneficial [4]. The researchers hypothesized that cannabis may preserve subcortical regions with high CB1 receptor density, including structures involved in memory, emotion, and motor control [4].

A systematic meta-review found that when abstinence periods of 72 hours or more were enforced before testing, no significant cognitive differences between heavy cannabis users and non-users remained [13]. This suggests that much of the measured cognitive impact may reflect acute intoxication and withdrawal rather than lasting impairment.

Critics of the deficit-focused research point to methodological concerns: many studies fail to control for tobacco and alcohol co-use, socioeconomic status, or pre-existing mental health conditions [7]. One 2022 study found that structural brain differences previously attributed to cannabis "reflected a history of polysubstance use" rather than cannabis alone [7].

The Funding and Conflict-of-Interest Landscape

Research funding in this area flows through channels with built-in tensions. The National Institute on Drug Abuse (NIDA), the primary funder of cannabis research in the United States, has faced criticism for what some researchers describe as a structural bias toward studying harms [16]. The Multidisciplinary Association for Psychedelic Studies (MAPS) has reported that NIDA rejected two FDA-approved medical marijuana research protocols, preventing the studies from proceeding [16]. NIDA has historically held a monopoly on the supply of cannabis available for federally approved research, further constraining the scope of inquiry [16].

On the other side, the expanding legal cannabis industry — projected to exceed $50 billion in U.S. sales — has its own incentive to fund or publicize research showing benign or positive outcomes. The Guha et al. study did not list its funding sources in the published press release [4]. The UK Biobank Mendelian randomisation study, conducted by Oxford University researchers, was funded through standard academic channels with no reported cannabis industry ties [2].

Most of the major studies discussed here — the ABCD Study, the JAMA Network Open study — received NIDA or NIH funding [3][8]. This is largely unavoidable given NIDA's dominant role, but readers should understand the institutional context.

Policy Meets Evidence

On December 18, 2025, President Trump issued Executive Order 14370 directing federal agencies to accelerate reclassification of marijuana from Schedule I to Schedule III under the Controlled Substances Act, while also expanding federal research into medical marijuana and CBD [17]. This represents a significant policy shift, though reclassification remains incomplete as of March 2026.

At the state level, cannabis legalization continues to expand with limited acknowledgment of neuroimaging research. A February 2025 Canadian study of 13.6 million individuals found a 270% increase in hospital care for cannabis use disorder following legalization, with schizophrenia diagnoses among those with prior cannabis use disorder rising from 7% to 16% [18]. Some experts have advocated for potency regulations modeled on European approaches — Switzerland caps THC content at 1%, France at 0.3% — but no U.S. state has implemented comparable limits [18].

No state regulatory body has cited the neuroimaging literature discussed here as a basis for policy changes. The research remains largely confined to academic journals and has not yet been translated into regulatory frameworks.

What Medical Cannabis Patients Stand to Lose

For medical cannabis patients, the stakes of this research are personal. PTSD is a qualifying condition for medical cannabis in nearly every state program, and it is the third most common condition cited by patients seeking authorization, behind chronic pain and anxiety [19]. An analysis of over 61,000 patients across 12 states confirmed the scale of medical reliance on cannabis [19].

Some clinical evidence supports this use: a study cited by the VA's National Center for PTSD found that participants using THC-dominant cannabis reported greater reduction in PTSD symptom severity over time, with cannabis users showing more than twice the rate of remission from PTSD diagnosis compared to controls [19]. Both houses of Congress have passed H.R. 2577, which would allow VA physicians to recommend cannabis to veterans in legal states [20].

Patient advocates argue that restricting access based on structural neuroimaging findings — particularly those that have not been shown to be causal — would disproportionately harm veterans, chronic pain patients, and others who have exhausted conventional treatment options. The VA's own research review characterized the evidence base for cannabis harms in this population as "limited" [19].

"The story is nuanced," Guha told the University of Colorado Anschutz newsroom. "It's not a case of cannabis being all good or all bad" [4].

What We Don't Know

Several gaps in the evidence deserve explicit acknowledgment:

• Potency: Most large studies, including the UK Biobank analyses, could not differentiate between THC and CBD content or account for the dramatic increase in cannabis potency over recent decades [4][18]. Today's products bear little pharmacological resemblance to those consumed by the older adults in biobank cohorts.

• Dose-response thresholds: No study has established a clear threshold — in grams, frequency, or duration — at which detectable structural changes appear. The JAMA Network Open study used a rough cutoff of 1,000 lifetime uses for "heavy" users but acknowledged this as imprecise [3].

• Long-term recovery data: While 72-hour to one-month abstinence data is available, rigorous longitudinal follow-up at 1 year, 5 years, and beyond post-cessation remains scarce, particularly with neuroimaging endpoints rather than cognitive testing alone [12][13].

• Demographic representation: The UK Biobank studies relied on predominantly White, relatively healthy British populations [6]. Generalizability to younger, more diverse populations with higher-potency products is uncertain.

The research to date suggests that cannabis affects the brain differently depending on when you start, how much you use, what else you consume alongside it, and possibly your genetic profile. What it does not yet establish — despite decades of study and over 29,000 published papers — is whether those effects are caused by cannabis, whether they matter clinically, or whether they should change how the substance is regulated. For the millions of Americans who use cannabis daily, and the policymakers deciding its legal future, the honest answer remains: we are still finding out.