Prenatal cannabis exposure linked to subtle brain and behavioral differences in children

Researchers mapped brain networks in more than 10,000 children to examine whether prenatal cannabis exposure leaves a detectable signature across cognition, behavior and mental health.

Want to read later? Download your clean, ad-free, print-friendly PDF copy by clicking here.

A large study published as an 'Article in Press' in the journal Translational Psychiatry found that prenatal cannabis exposure (PCE) was associated with differences in brain functional connectivity in late childhood, along with higher behavioral and emotional problem scores and lower cognitive performance in several areas.

The study also found partial overlap between the connectivity patterns associated with PCE and those linked with psychopathology and cognitive performance, identifying candidate neurobiological pathways for future investigation rather than established developmental mechanisms.

Background

Perceptions of cannabis as relatively safe during pregnancy have become increasingly common, and recent US data indicate that cannabis use during pregnancy has increased. But little is known about how PCE is linked to offspring neurodevelopment.

Previous studies have reported associations with cognitive, behavioral, and brain outcomes, but findings have been inconsistent. In response, public health agencies recommend that cannabis use be avoided by pregnant and lactating women, and by women who plan to become pregnant.

The prenatal period is crucial to brain and behavioral development. The endocannabinoid system (ECS) in the fetal brain is active from the early stages of fetal development. This coordinates the various neurobiological processes involved in the development of cognitive, emotional, and reward circuits.

Cannabis contains the highly psychoactive substance Δ9-tetrahydrocannabinol (Δ9-THC), which readily crosses the placental barrier and binds to the cannabinoid receptors CB1 and CB2, both part of the ECS. PCE could disrupt ECS-mediated regulation of fetal brain maturation.

Human and preclinical studies have reported neurochemical and molecular changes associated with PCE that could be relevant to later neuropsychiatric vulnerability. Δ9-THC may reduce serotonin levels in the brain, which has been associated with a higher risk of mood and anxiety disorders.

Previous studies have associated PCE with differences in neurodevelopmental outcomes, including cognition, behavior, and emotional regulation. Externalizing symptoms and poorer mental health are also more common in later childhood and adolescence in such individuals.

Advanced neuroimaging, including functional magnetic resonance imaging (fMRI) and functional connectivity analyses, has shown PCE-associated differences across large-scale neurocognitive systems in the brain that support executive control, emotional regulation, and reward processing. These domains are particularly sensitive to disruption during neurodevelopment.

Taken together, the evidence suggests that PCE may perturb the maturation of brain circuits important for affective and cognitive processing. The current study aimed to integrate PCE-associated behavioral and brain connectivity patterns in children to identify candidate neurobiological pathways for future research.

Study characteristics

Researchers analyzed baseline data from the Adolescent Brain Cognitive Development (ABCD) Study, focusing on children with complete behavioral and neuroimaging assessments. The mean participant age was about 10 years.

Overall, 10,838 children had at least one high-quality resting-state fMRI scan, including 751 classified as having PCE and 10,087 without reported exposure.

PCE was determined from maternal reports of marijuana use either before the mother knew she was pregnant or after she became aware of the pregnancy. Prenatal alcohol and nicotine exposure were also assessed and included as covariates. Children who reported cannabis use at baseline were excluded.

The researchers used the NeuroMark approach to identify 53 intrinsic connectivity networks (ICNs) spanning seven functional domains. Whole-brain functional network connectivity (FNC) was calculated from correlations between ICN time courses, and the researchers then tested its association with PCE. They then examined relationships between FNC and behavioral and cognitive measures.

PCE associated with children’s mental health and cognitive functioning

Children with PCE had higher CBCL scores, but effect sizes were small. The largest difference was for the Rule-Breaking Behavior scale, and the smallest for the Somatic Problem scale. This indicates increased behavioral and emotional problems in association with PCE, with the largest differences involving rule-breaking and other behavioral problems.

PCE was also associated with lower cognitive performance, with the most pronounced effect on the Crystallized Cognition score. This indicates worse performance on tasks requiring acquired knowledge and verbal comprehension. Conversely, Fluid Cognition scores were modestly higher among children with PCE.

PCE associations with FNC

The FNC analysis showed widespread differences in functional connectivity among children with PCE. Exposed children showed increased connectivity within subcortical and sensorimotor networks, and between default-mode and sensory/subcortical networks.

FNC was lower within the default-mode network, between subcortical and sensory networks, and between default-mode and cognitive-control networks. Some connections between the cognitive-control and other networks showed a mixed pattern.

Overlap of connectivity alterations with cognitive and mental health

The researchers next asked whether the brain-network patterns associated with PCE resembled those linked to children's behavioral and cognitive outcomes. Analyses linking FNC with CBCL scores identified patterns that partly overlapped with those associated with PCE. Increased FNC in the subcortical and sensorimotor networks, and between the default-mode and sensory domains, was linked to higher behavioral and emotional problem scores, as was decreased FNC between subcortical and sensory domains.

Connectivity involving the cognitive-control domain showed mixed patterns of association between FNC and CBCL scores.

Increased connectivity within the cerebellum and between the cerebellar and visual networks was linked to higher CBCL problem scores, but also to reduced connectivity between the CB and multiple domains outside the cerebellum.

The strongest spatial correspondence was with rule-breaking behavior, followed by conduct problems, aggressive behavior, and social problems.

The FNC-cognitive performance association analyses showed patterns that were opposite in direction to the PCE-FNC associations, fr thesures that were lower among children with PCE. In practical terms, better performance on these measures was generally linked to connectivity patterns running in the opposite direction to those associated with PCE.

Reduced FNC in the subcortical and sensorimotor domains, and between the default-mode and sensory networks, was associated with higher cognitive performance. Conversely, increased FNC between the subcortical and sensory domains was linked to higher cognitive performance. Mixed results were obtained for connectivity involving the cognitive-control domain.

Reduced connectivity within the cerebellum, and between cerebellar and visual domains, was linked to greater cognitive performance, as was higher connectivity between the CB and other functional domains.

The strongest negative spatial correlations between PCE-associated and cognition-associated connectivity maps were observed for picture vocabulary, crystallized cognition, oral reading recognition, and list-sorting working memory. Fluid cognition, the cognitive measure that was higher among exposed children, was positively correlated with the PCE-related FNC.

These findings and the small effect sizes are concordant with earlier studies. The authors suggest that “small effects may be an inherent characteristic of complex neurodevelopmental and behavioral phenotypes rather than a limitation unique to the present study.”

The researchers found that cerebellar connectivity was associated with cognitive and behavioral variation but not with PCE. This suggests that, despite evidence that the cerebellum contributes to cognition and psychopathology, it may not be the major pathway linking PCE with childhood behavioral outcomes.

The authors identify disruption of developing endocannabinoid signaling by Δ9-THC as one biologically plausible interpretation of the observed associations.

Limitations

The effect sizes observed were small, so the associations should not be interpreted as clinically meaningful at the individual level. They also do not support diagnostic or predictive use of these FNC patterns.

The study did not examine structural brain alterations, though these may also be associated with PCE. The observational nature of the study prevents causal inferences, while the cross-sectional data used here cannot distinguish whether these neurobehavioral changes reflect temporary adaptive alterations or stable neurodevelopmental trajectories. Future studies could use longitudinal data from the ABCD cohort to track the evolution of functional connectivity and cognitive-behavioral outcomes in relation to PCE.

Further, PCE is often linked to other factors, including socioeconomic deprivation and psychosocial stressors, contributing to residual confounding. Sensitivity analyses adjusting for household income, caregiver education, family history of substance-use-related problems, birth weight, and additional prenatal substance exposures produced largely unchanged results, but could not remove all residual confounding. The study contained few children exposed to cannabis alone, and the findings should not be interpreted as evidence of the isolated effects of cannabis exposure. Retrospective caregiver reports of PCE may also have introduced recall and social desirability bias.

Conclusions

“Collectively, these findings indicate that PCE is associated with widespread differences in functional connectivity, some of which overlap with patterns related to psychopathology and cognitive performance.” These are population-level rather than individual-level patterns, but their directionality and causality remain to be established.

The convergence of patterns across multiple domains suggested differences in subcortical-cortical integration, a process important for cognition and emotional regulation. The authors caution that these findings do not indicate causal or mediating effects, but identify candidate neurobiological pathways for further investigation.

The authors propose that future longitudinal studies combining multi-modal measures, including repeated imaging, behavioral and cognitive assessments, and biological exposure measures, could clarify the developmental significance of the associations.

Journal reference:

• Fu, Z., Hutchison, K., Guha, A., & Calhoun, V. D. (2026). Prenatal cannabis exposure associated with altered brain connectivity: Neural correlates of cognitive and mental health variability in offspring. Translational Psychiatry. DOI: 10.1038/s41398-026-04478-5, https://www.nature.com/articles/s41398-026-04478-5