Chronic cannabis use differentially modulates neural oscillations serving the manipulate versus maintain components of working memory processing

Abstract

The legalization of recreational cannabis use has expanded the availability of this psychoactive substance in the United States. Research has shown that chronic cannabis use is associated with altered working memory function, however, the brain areas and neural dynamics underlying these affects remain poorly understood. In this study, we leveraged magnetoencephalography (MEG) to investigate neurophysiological activity in 45 participants (22 heavy cannabis users) during a numerical WM task, whereby participants were asked to either maintain or manipulate (i.e., rearrange in ascending order) a group of visually presented numbers. Significant oscillatory responses were imaged using a beamformer and subjected to whole-brain ANOVAs. Notably, we found that cannabis users exhibited significantly weaker alpha oscillations in superior parietal, occipital, and other regions during the encoding phase relative to nonusers. Interestingly, during the maintenance phase, there was a group-by-condition interaction in the right inferior frontal gyrus, left prefrontal, parietal, and other regions, such that cannabis users exhibited weaker alpha and beta oscillations relative to nonusers during maintain trials. Additionally, chronic cannabis users exhibited stronger alpha and beta maintenance responses in these same brain regions and prolonged reaction times during manipulate relative to maintain trials, while no such differences were found in nonusers. Neurobehavioral relationships were also detected in the prefrontal cortices of nonusers, but not cannabis users. In sum, chronic cannabis users exhibit weaker neural oscillations during working memory encoding but may compensate for these deficiencies through stronger oscillatory responses during memory maintenance, especially during strenuous tasks such as manipulating the to-be remembered items.

Keywords: MEG; Magnetoencephalography; Marijuana; Short-term memory; Substance use; cannabis use disorder.

Fig. 1.

Experimental paradigm and behavioral results. (a) Each trial of the numerical working memory manipulation (WMM) task began with a fixation cross embedded in the middle of an empty 4 × 1 grid for 1.8 ± 0.2 s (fixation), which served as the baseline period. Four single numbers (1–9) then appeared within the grid and remained for 1.2 s (encoding). The digits then disappeared and simultaneously the grid changed color (lighter/darker), indicating whether the participant needed to maintain the numbers in original order (maintain) or rearrange them into numerically ascending order (manipulate) during the 2.5 s maintenance period. A single probe digit was then presented in one of the four boxes for 1.6 s (retrieval) and participants needed to respond with a button press whether the number was in the correct box (correct: index, incorrect: middle finger). (b) Example of correct trials for the maintain and manipulate conditions. (c) Performance on the numerical WMM task. The top plot shows the significant main effect of condition and the group-by-condition interaction effect (p < .05), which indicated that cannabis users responded slower during manipulate relative to maintain trials, while nonusers did not show such differences. The bottom plot shows a significant main effect of condition on accuracy across all participants. Each violin plot contains the individual data points and mean (x). *p < .05; ***p < .001.

Fig. 2.

Group differences in alpha oscillatory activity during encoding. Significantly stronger alpha oscillations were observed in the left inferior parietal cortices, right superior parietal lobule, and the left lateral superior occipital cortices during early encoding and left posterior cingulate during late encoding in nonusers relative to cannabis users. Each violin plot contains the individual data points and mean (x). IPC: inferior parietal cortices; SPL: superior parietal lobule; LOC: lateral occipital cortices; PCC: posterior cingulate cortices. *** p < .0005.

Fig. 3.

Alpha and beta task condition by group interaction effects. Whole brain 2 × 2 ANOVAs were performed on alpha and beta beamformer images during early maintenance. Significant clusters were defined at the p < .0005. (a) Significant group by task condition differences were found in the right prefrontal cortices, right intraparietal sulcus, left inferior temporal, and left posterior cingulate for alpha oscillations. Follow-up testing indicated significantly stronger alpha oscillations (i.e., more negative relative to baseline) in nonusers relative to cannabis users during the maintenance condition, as well as stronger alpha responses during manipulate compared to maintain trials only in the user group. (b) Similar group by task condition differences were found in the beta range within the left DLPFC and the right parietal, with the group maintenance condition difference being significant in the left DLPFC and the conditional difference in cannabis users being significant in both regions. PFC: prefrontal cortices; ITG: inferior temporal gyrus; IPS: intraparietal sulcus; PCC: posterior cingulate cortex; DLPFC: dorsolateral prefrontal cortices. * p < .05, ** p < .005, *** p < .001.

Fig. 4.

Task condition main effects. Significantly stronger alpha and beta oscillations were observed in the manipulate relative to maintain condition during early maintenance. The relevant regions are indicated by circles with arrows linking to their respective violin plots. TPJ: temporo-parietal junction, MFG: middle frontal gyrus. *** p < .0005.

Fig. 5.

Neurobehavioral relationships. (a) Stronger alpha oscillations in the right prefrontal cortices were significantly associated with better accuracy only in the nonuser group during maintain trials. (b) A similar relationship was found between accuracy and beta oscillations in the left DLPFC, such that stronger beta was associated with better accuracy only in the nonusers, although it was during both manipulate and maintain trials. *p < .05, **p <. 01.