Heavy cannabis use linked to worse cognition in young adults, but mild use may help those with HIV

The findings from this study arrive at a moment when cannabis policy across the United States is shifting faster than the science needed to guide it. State-level legalization has expanded access dramatically, yet the clinical literature on how cannabis affects the developing brain remains fragmented, with studies differing widely in how they define exposure, which cognitive domains they measure, and which populations they sample. One recurring challenge is that much of the existing research treats cannabis use as a binary—users versus non-users—which obscures the possibility that dose and frequency matter in fundamentally different ways. By categorizing use into none, mild to moderate, and heavy tiers based on days of use per week, and by examining use across three overlapping recall windows of 30, 180, and 365 days, the current study adds granularity that binary designs cannot provide. The consistency of results across the 30-, 180-, and 365-day windows is itself informative, suggesting that the associations observed are not artifacts of a single snapshot of recent behavior but reflect patterns of use sustained over the better part of a year.

The choice of cognitive domains was not arbitrary. The battery covered verbal memory, processing speed, working memory, executive functions, and overall fluid cognition—domains that developmental neuroscience identifies as continuing to mature through the mid-twenties. The prefrontal cortex, which supports executive functions such as inhibition, planning, and cognitive flexibility, undergoes protracted synaptic refinement well into the third decade of life, and the hippocampal systems supporting episodic memory are similarly sensitive to pharmacological and inflammatory perturbations during this window. This developmental context is what makes research in young adults aged 18 to 24 particularly consequential: exposures that might be tolerable in a fully mature brain may carry different weight in one that is still consolidating its final architecture. The finding that heavy use was associated with worse immediate and delayed word list recall aligns with a substantial literature implicating cannabinoid signaling in hippocampal memory function. Cannabinoid type 1 receptors are densely expressed in hippocampal pyramidal neurons, where endocannabinoid signaling normally acts as a brake on neurotransmitter release; exogenous THC engages these same receptors and can disrupt the synaptic plasticity mechanisms thought to underlie the encoding and consolidation of new memories.

The distinction between word list recall and story recall in the results deserves attention. These two measures both tap verbal memory, but they differ in structure: word list recall relies heavily on the ability to encode and retrieve discrete, unrelated items, whereas story recall engages additional integrative processes that bind narrative elements into a coherent whole. The observation that heavy use, relative to mild or moderate use, was associated with worse story recall, while heavy use relative to no use was associated with worse word list recall, hints that different aspects of the memory system may be differentially sensitive to cannabis exposure. This kind of dissociation is common in neuropsychological research and often points to partially separable neural substrates, though confirming such interpretations would require targeted neuroimaging or longitudinal designs that the present cross-sectional data cannot support.

The HIV dimension of the study introduces a genuinely novel angle. Even in the era of effective antiretroviral therapy, which suppresses peripheral viral replication to undetectable levels in most people who maintain treatment, HIV continues to exert effects on the central nervous system. The virus establishes reservoirs in immune cells that traffic into the brain early in infection, and even suppressed infection is associated with persistent immune activation. Elevated levels of inflammatory signaling molecules—such as interleukin-6 and soluble CD163, a marker of monocyte and macrophage activation—have been documented in people living with HIV and correlated with neurocognitive impairment in both adult and youth cohorts. The prevailing hypothesis is that this chronic, low-grade neuroinflammation, possibly seeded by early gut dysbiosis and sustained immune activation, gradually erodes cognitive function in the fluid domains that the study measured. Against this backdrop, the endocannabinoid system offers a plausible mechanistic counterweight. Cannabinoid receptors are abundant on immune cells, and their activation can dampen the release of pro-inflammatory cytokines. This immunomodulatory capacity is the theoretical basis for the intriguing—and admittedly preliminary—suggestion that mild to moderate cannabis use might be associated with better performance on some memory and executive measures among young adults with HIV.

It is worth emphasizing how unusual this pattern is in the broader cannabis literature. Most studies of cannabis and cognition in young people report either null effects or negative associations, with heavier use generally faring worse. The suggestion of a non-linear relationship in the HIV group—in which mild to moderate use was associated with better performance while heavy use remained deleterious—echoes a hormetic framework sometimes invoked in pharmacology, where low doses of a compound produce effects opposite in direction to high doses. In the context of chronic neuroinflammation, a modest anti-inflammatory signal could theoretically shift the balance toward neuroprotection, whereas the higher exposures associated with heavy use could overwhelm any benefit through direct neurotoxic or neurodevelopmental effects. The authors’ caution that causality cannot be determined from cross-sectional data is essential here. Several alternative explanations remain viable. Young adults with HIV who use cannabis mildly may differ from non-users in ways that also affect cognition—differences in sleep, pain, anxiety, medication adherence, or social circumstances could all confound the apparent protective association. Reverse causation is also possible: individuals experiencing fewer cognitive difficulties might feel more comfortable using cannabis moderately, rather than the other way around.

The statistical approach also merits explanation for readers less familiar with it. Generalized estimating equations are a regression framework well suited to repeated measures or clustered data, allowing the researchers to model cognitive outcomes while accounting for the correlation among related measures and adjusting simultaneously for HIV status, cannabis category, and their interaction. The interaction term is the critical test of whether cannabis effects differ by HIV status; its non-significance in the categorical analyses indicates that the study, powered for its modest sample size, could not statistically confirm that the cannabis-cognition relationship differs between groups. The exploratory continuous analyses, which treated frequency as a dose variable, were more suggestive of divergence—a linear dose-response in young adults without HIV, where more use tracked with worse performance, versus a non-linear pattern in those with HIV—but exploratory findings of this kind require replication before they can be considered established. With 38 young adults with HIV and 61 without, the study’s ability to detect interaction effects is limited, and null interaction results should not be read as evidence of equivalence.

The demographic composition of the sample raises further considerations. With roughly a third of participants identifying as White, the cohort reflects the demographic realities of HIV epidemiology in the United States, where Black and Hispanic communities bear a disproportionate share of new infections among young people. This composition enhances the relevance of the findings for the populations most affected by HIV, but it also means the results may not generalize uniformly across all demographic groups. Equally important, the authors note that differences in the proportion of females between the HIV and non-HIV groups prevented meaningful analysis of sex differences. This is a significant limitation given accumulating evidence that gonadal hormones interact with both cannabinoid signaling and inflammatory processes, and that sex may moderate the cognitive effects of cannabis. Future studies with balanced samples will be needed to determine whether the patterns observed here hold for young men and young women alike.

The finding that young adults with HIV performed worse than their uninfected peers across fluid cognition, memory, processing speed, and executive functioning is consistent with a large body of research documenting HIV-associated neurocognitive effects even in virally suppressed individuals. What this study adds is the demonstration that these deficits coexist with, and may interact with, modifiable behavioral factors like cannabis use. From a clinical standpoint, this matters because cannabis is already widely used among people living with HIV, often self-directed toward managing pain, sleep disturbance, and anxiety. If frequency of use genuinely differentiates cognitive benefit from harm in this population, then harm-reduction conversations in HIV care settings could become more nuanced—moving beyond simple abstinence messaging toward guidance about patterns of use, pending stronger evidence.

The product composition issue looms large over any translation of these findings. Cannabis available today in legal markets differs substantially in cannabinoid profile from the material on which much of the older literature was based. Tetrahydrocannabinol concentrations have risen markedly over recent decades, while cannabidiol content has often declined in high-THC products. Because THC and CBD have distinct and sometimes opposing pharmacological and immunological profiles—THC being primarily psychoactive and CBD largely non-intoxicating with notable anti-inflammatory signaling—the ratio of these compounds could plausibly determine whether a given pattern of use helps or harms an inflamed nervous system. The present study could not disentangle these constituents, since frequency of use days does not capture potency, dose per occasion, or route of administration, all of which shape the pharmacological exposure actually delivered to the brain.

Looking forward, the most valuable next step would be a longitudinal design that follows young adults with and without HIV over time, measuring cannabis use, inflammatory biomarkers, and cognitive performance repeatedly. Such a study could begin to disentangle the temporal ordering of the associations observed here and test the neuroinflammatory hypothesis directly by examining whether changes in cytokine levels or immune activation markers mediate the relationship between use frequency and cognition. Randomized or quasi-experimental approaches examining specific cannabinoid compounds at controlled doses would provide even stronger causal evidence, though such trials in young adults raise ethical and regulatory complexities. In the meantime, the study’s central contribution is conceptual: it reframes cannabis effects not as uniformly good or bad, but as potentially dependent on the biological context of the user—in this case, the presence or absence of a chronic neuroinflammatory condition—and on the frequency with which the substance is used. That reframing, if validated by future work, could reshape how clinicians, researchers, and policymakers think about cannabis guidance for the growing population of young people living with HIV.

Subject of Research: Cannabis use frequency may differentially impact cognitive performance in young adults with and without HIV

Article Title: Cannabis use frequency may differentially impact cognitive performance in young adults with and without HIV

Article References: Nichols, S. L., Yang, K., Staley, D. A., O'Keeffe, J., Tu, X. M., Happer, J. P., Hall, K., Deng, M., Hasheem, S., Wang, R., Lansing, A. E., Jacobus, J., & Fennema-Notestine, C. (2026). Cannabis use frequency may differentially impact cognitive performance in young adults with and without HIV. Frontiers in Adolescent Medicine, 4, Article 1716084. https://doi.org/10.3389/fradm.2026.1716084

Image Credits: AI Generated

DOI: 10.3389/fradm.2026.1716084

Keywords: Cannabis, frequency, differentially, impact, cognitive, performance, young, adults, without, scientific research