rethinkTHC Search
Menu
Study breakdown

Higher Levels of One Brain Chemical May Protect Memory in Heavy Cannabis Users

Cross SectionalModerate evidence
The takeaway

Among heavy cannabis-using young adults, those with higher blood levels of the endocannabinoid 2-AG showed better verbal memory, while higher AEA levels were associated with worse performance — suggesting endocannabinoid profiles may predict who is most vulnerable to cannabis-related cognitive effects.

Neuroscience researchers, addiction medicine specialists, and young adult cannabis users interested in cognitive risk factors.

What the researchers found

Heavy cannabis use was associated with worse verbal memory. However, significant interactions showed that heavy users with elevated 2-AG concentrations performed better on verbal learning and memory, while heavy users with higher AEA concentrations performed worse. In men specifically, high 2-AG with heavy use predicted better verbal learning.

Why it matters

Not all heavy cannabis users experience the same cognitive effects. This study suggests that individual differences in endocannabinoid levels may explain why some heavy users maintain good memory while others don't — potentially opening the door to biomarker-based risk assessment.

The numbers in context

87 participants (ages 18–20, 63% female). Three cannabis groups: no use, light (<8 days/month), heavy (≥8 days/month). High 2-AG + heavy use = better memory. High AEA + heavy use = worse learning. Three-way interaction: men with heavy use and high 2-AG had best verbal learning.

How the study worked

87 participants aged 18–20 (63% female) completed measures of past-30-day cannabis use, verbal learning and memory (RAVLT), and blood draws for serum endocannabinoid analysis (2-AG and AEA). Linear regressions examined cannabis group × endocannabinoid interactions on memory, with sex as a moderator.

What this study cannot tell us

Cross-sectional — cannot determine if endocannabinoid levels are a cause or consequence of cannabis use patterns. Small sample. Serum levels may not reflect brain endocannabinoid concentrations. Only one cognitive domain tested (verbal memory).

How to read the evidence

Cross-sectional study with biological markers and validated cognitive testing, but small sample and cross-sectional design limit causal inference.

When this study was published

Published 2025.

The bigger picture

The opposing roles of 2-AG and AEA in cannabis-related memory performance suggest the endocannabinoid system's response to regular cannabis use is complex. 2-AG may compensate for THC's effects while elevated AEA may indicate stress or dysregulation. This could eventually inform personalized risk assessment.

Questions still open

  • Could endocannabinoid profiling identify which young cannabis users are most at risk for cognitive problems? Does 2-AG rise as a compensatory mechanism during regular use? Would targeting AEA levels protect memory in heavy users?

Common questions

Can a blood test predict if cannabis will affect my memory?
This early research suggests endocannabinoid levels (2-AG and AEA) may one day help predict cognitive vulnerability. However, this is a single small study and clinical biomarker testing for cannabis risk is not yet available.
Why do some heavy users keep sharp memory while others don't?
This study suggests individual differences in endocannabinoid system function may explain the variation. Higher 2-AG may provide a protective buffer, while higher AEA may indicate the brain is struggling to adapt to regular cannabis exposure.

Read the original research

Young Adult Cannabis Use and Circulating Endocannabinoid Concentrations on Cognitive Performance.

Frontiers in adolescent medicine, 3

Citation

Wallace, Alexander L; Baca, Rachel; Andrade, Gianna; Hillard, Cecilia J; Happer, Joseph P; Wade, Natasha E; Courtney, Kelly E; Mejia, Margie Hernandez; Jacobus, Joanna. (2025). Young Adult Cannabis Use and Circulating Endocannabinoid Concentrations on Cognitive Performance.. Frontiers in adolescent medicine, 3. https://doi.org/10.3389/fradm.2025.1538448

Explore the wider topic