THC-Related Picture-Memory Differences Two Days Later Do Not Prove Continuous Intoxication
A controlled human study found poorer recognition of pictures two days after participants had viewed them following oral THC. That result concerns memory for information encountered during drug exposure. It does not establish that participants remained intoxicated, or unable to drive, throughout the intervening two days.
The drug was given before learning, not at the later test
Ballard, Gallo and de Wit enrolled 25 healthy volunteers in a double-blind, placebo-controlled THC crossover experiment. On separate visits, participants received placebo, 7.5 mg dronabinol or 15 mg dronabinol. They viewed pictures 90 minutes after taking the capsule, then returned two days later for recognition testing without receiving a study drug.
The pictures included pleasant, neutral and unpleasant scenes. At the later test, 60 previously viewed pictures were mixed with 60 unfamiliar pictures. Participants judged which ones they had seen. The main accuracy measure subtracted false alarms—calling an unfamiliar picture familiar—from correctly recognized pictures.
One THC participant’s memory data were excluded for not following task instructions, leaving 24 participants in those analyses. A separate 25-person dextroamphetamine experiment appeared in the same paper. Participants did not receive an amphetamine–THC combination, and its results cannot be presented as a combined-drug interaction trial.
What the significant and null findings show
THC significantly reduced overall recognition accuracy, with a reported dose effect of p = .019; both tested doses differed from placebo in follow-up comparisons. For pleasant pictures, accuracy was significantly lower (p = .024). The unpleasant-picture accuracy result did not meet the study’s .05 threshold (p = .081), and neutral-picture accuracy was not significantly changed.
The formal dose-by-picture-valence interaction for accuracy was nonsignificant. Consequently, a significant result in one picture category and a nonsignificant result in another should not be treated as proof that the drug’s effects differed between those categories. The pattern suggested an emotional-memory question, but the interaction did not establish that distinction.
THC reduced correct recognition of previously seen pictures overall, while false-alarm rates were not significantly changed. The study also did not demonstrate the hypothesized bias toward remembering positive rather than negative material. These outcomes describe different aspects of memory and should remain separate.
A delayed memory effect is different from ongoing intoxication
Learning, retaining information and retrieving it later are different stages. Here, the investigators administered THC before picture viewing and tested recognition later under drug-free study conditions. The design therefore addresses memory formation, including encoding and potentially early consolidation. It does not isolate encoding from consolidation.
A person could perform differently on a later test because the original information was learned differently. That possibility does not require a new drug effect at the time of testing. Conversely, the result does not establish that every cognitive function had recovered by that visit. This experiment did not continuously measure impairment over 48 hours or conduct an on-road driving assessment.
The reported biological screening involved urine drug tests and breath-alcohol checks for visit compliance. The study did not provide quantitative THC concentrations in whole blood, plasma, serum or oral fluid with which to define an individual impairment threshold. Capsule dose and a later recognition score cannot be converted into such a concentration.
How far the findings can be applied
This was human research using oral synthetic THC, rather than an animal or cell experiment or a test of smoked whole-plant cannabis. Participants were healthy adults aged 18–35 with previous cannabis experience; frequent use above three times weekly was excluded. The small, selected sample limits generalization to other populations, doses and routes.
Placebo recognition was already high, potentially limiting the ability to detect some differences. The analysis also used directional testing and exploratory comparisons. Laboratory picture recognition does not establish the accuracy of a particular witness’s account of a stressful real event.
For a memory-related opinion, identify when exposure occurred relative to learning and questioning, which error was measured, and whether the comparison involved the same task and population. Our analysis of false alarms versus forgetting in cannabis recognition research explains why naming the particular error matters.
Primary research
Ballard ME, Gallo DA, de Wit H. Pre-encoding administration of amphetamine or THC preferentially modulates emotional memory in humans. Psychopharmacology. 2013;226:515–529.
