Remembering Molly: Immediate and delayed false memory formation after acute MDMA exposure

MDMA (3,4-methylenedioxy-methamphetamine), commonly known as "Molly" or ecstasy, is an entactogenic phenethylamine that releases and inhibits reuptake of monoamines including serotonin. While clinical interest in MDMA is rising—most notably as an adjunct in psychotherapy for PTSD—acute intoxication has been associated with selective cognitive effects: some domains remain intact or improve slightly (for example attention), whereas verbal and working memory are often vulnerable. Prior controlled studies have reported small, transient impairments of verbal memory after a single 75 mg dose, and one recent experiment suggested MDMA may alter recollection versus familiarity and show a trend toward increased false alarms for positive stimuli. Given MDMA's serotonergic profile, its memory effects might resemble those of psychedelics more than classic stimulants. Kloft and colleagues designed a double-blind, placebo-controlled study to examine whether acute MDMA (75 mg) alters susceptibility to false memory. The investigators tested 60 healthy volunteers with prior MDMA exposure on three established false-memory paradigms: the Deese/Roediger-McDermott (DRM) associative word-list task and two virtual-reality (VR) misinformation tasks modelled on eyewitness and perpetrator scenarios. Memory was assessed both during acute intoxication (immediate test) and approximately 7 days later when sober (delayed test). The primary expectation was that MDMA would impair true memory and potentially increase false-memory rates compared with placebo.

The study used a randomized, placebo-controlled, double-blind mixed design with Group (MDMA vs placebo) as a between-subjects factor and Time (immediate vs delayed) as a within-subjects factor. Participants were matched across groups for age, sex and education, and randomisation sequences counterbalanced drug order and VR scenario assignment. The protocol received ethical approval and the trial was registered. Sixty-one healthy volunteers with prior MDMA experience were enrolled; 60 completed all procedures (28 female, 33 male, mean age 23.0, SD 3.3, range 18–32). Participants were light-to-moderate MDMA users (lifetime occasions reported between 3 and 60) to comply with local ethics. Screening included medical history, physical examination, blood and urine tests, and ECG. Subjects abstained from drugs for 7 days and from alcohol for 24 h before each test day; drug and alcohol tests were applied on test days. Each participant received 75 mg oral MDMA and placebo on separate test days spaced 7 ± 1 days apart. Memory testing occurred during peak drug effects (approximately 60–120 min post-dose; immediate condition) and at a 1-week sober follow-up (delayed condition). A venous blood sample taken 90 min post-dose quantified MDMA and MDA concentrations. Subjective intoxication was rated on 100 mm visual analogue scales. The DRM task assessed spontaneous associative false memory. On each test day participants studied 15 lists of ten related words, then completed filler tasks and a recognition test. Test items included previously presented words (old), critical lures (strongly associated but not presented), related lures (moderately associated), and unrelated lures. Two parallel DRM versions were used and each participant completed four test instances (immediate and delayed for each drug condition). The misinformation paradigm comprised two fully immersive VR scenarios (eyewitness on a train, and perpetrator in a bar), counterbalanced across sessions. Misinformation was introduced either via a virtual co-witness (eyewitness) or via leading questions at test (perpetrator). Recognition interviews were administered 30 min after encoding (immediate) and at 7-day follow-up (delayed) and contained presented, suggested and non-suggested items. Primary outcome measures were true memory (hit rate, proportion correctly recognised) and false memory (false alarm rate by lure type). Signal detection measures—sensitivity (d') and response bias (c)—were computed. Statistical analyses comprised repeated-measures ANOVAs: for DRM, 2 (Drug: MDMA vs placebo) x 4 (Level of association: old, critical, related, unrelated) within-subjects models; for misinformation tasks, 2 (Group) x 3 (presented, suggested, non-suggested) between-subjects ANOVAs at each time point. Pairwise t-tests were used for signal detection parameters. When sphericity was violated Greenhouse-Geisser corrections were applied. Equivalence testing was also conducted for null findings using a smallest effect size of interest (SESOI) set to Cohen's d = 0.27 (the smallest DRM effect observed). Additional analyses exploring VR telepresence and item novelty/repetition were reported in the Supplementary Information.

Sixty participants completed the protocol. Measured MDMA serum concentrations 90 min post-dose ranged from 51.2 to 287.9 ng/ml (mean 147.9, SD 46.0), and MDA ranged from 0.0 to 137.3 ng/ml (mean 12.1, SD 23.5). MDMA reliably increased subjective ratings of intoxication (all p's < 0.001, effect size ω2 = 0.56–0.58). DRM performance showed the expected gradient with highest recognition for old words and decreasing recognition for lures of lower associative strength. Statistically, a Drug x Level of association interaction was detected at immediate testing, indicating that MDMA affected some word categories differently than others. Specifically, during acute intoxication MDMA increased false alarms for related (moderately associated) lures on the immediate DRM test. True recognition (hit rates) was modestly impaired by MDMA both at immediate testing and at the 1-week delayed test. At delayed testing, MDMA was associated with reduced false recognition for critical (highly associated) lures. Signal detection analyses indicated reduced sensitivity (d') under intoxication in the DRM, while response bias (c) was generally unchanged by MDMA in these tests. In the eyewitness VR scenario there were no significant Group x Level effects or main effects of Group at the immediate test; true recognition rates were high and suggestion-based false memory rates low, with similar performance in MDMA and placebo. At the delayed test a main effect of Group emerged: the placebo group showed higher overall recognition [F(1,57) = 7.88, p = .007, ω2 = 0.06], driven by elevated suggestion-based false memories [F(1,57) = 8.68, p = .005, Cohen's d = 0.77]. Signal detection analyses showed a group difference in response bias at follow-up [t(57) = -2.70, p = .009, Cohen's d = 0.70], with the MDMA group responding more conservatively; sensitivity did not differ significantly. For the perpetrator VR scenario no significant group interactions or main effects were observed at the immediate test. At delayed test there was no statistically significant Group main effect, although p = .052 was noted (not meeting conventional significance). Signal detection parameters did not differ between groups for this scenario. Because several misinformation analyses yielded null results, equivalence tests were performed for eyewitness immediate, perpetrator immediate and perpetrator delayed conditions using a SESOI of d = 0.27. In all cases at least one one-sided test had p > 0.05, so statistical equivalence between MDMA and placebo could not be concluded. Additional analyses in the Supplementary Information indicated that MDMA increased subjective VR presence but this did not act as a significant covariate; isolating novel versus repeated test items suggested the DRM effects were most robust for true memory and did not materially change findings in the misinformation paradigms.

Kloft and colleagues present the first controlled study examining acute MDMA effects on false-memory formation using both an associative word-list (DRM) and applied misinformation paradigms. The principal finding is a complex, modest memory profile: small impairments of true verbal recognition were observed in the DRM during intoxication and at 1-week follow-up, MDMA elevated false alarms for related lures during the immediate test, and reduced false recognition for highly associated critical lures at delayed testing. By contrast, MDMA did not consistently increase susceptibility to suggestion in the two VR-based misinformation tasks; when group differences emerged at follow-up they reflected higher suggestion-based false memories in the placebo group and a more conservative response bias in the MDMA group. The investigators interpret the pattern as compatible with an encoding deficit account: if MDMA impairs initial processing and storage of item-level and gist information, associative or gist-based false memories may be differentially affected depending on whether intoxication is present at encoding or retrieval. This explanation aligns with theories of false memory that emphasise gist or associative activation as drivers of false recognitions. The authors also note parallels with research on cannabis/THC and with prior findings that MDMA can alter recollection versus familiarity, while emphasising that the observed effects were generally small and sometimes inconsistent. Signal detection results suggest reduced discriminability under MDMA but no overall shift toward a more liberal response bias; in one instance participants on MDMA were more conservative at follow-up. Several limitations and uncertainties are acknowledged. The study design did not fully disentangle encoding from retrieval processes, recognition tests may be less sensitive to drug effects than recall, and the 75 mg dose—though clinically and recreationally relevant—may not capture dose-dependent effects. Equivalence testing could not rule out small effects in some misinformation outcomes. The authors also point out that the within-subjects DRM procedure yields greater statistical power than the between-groups misinformation comparisons, which could contribute to differential sensitivity across tasks. Practical implications are discussed cautiously: while MDMA impaired true recognition modestly, the small effect sizes and mixed pattern preclude strong claims about forensic or clinical consequences. The researchers recommend further work that separates memory phases, uses recall measures as well as recognition, examines higher doses and differing pharmacodynamic windows, and explores suggestibility via imagination-inflation paradigms. They also highlight the relevance of this research for MDMA-assisted psychotherapy, where false-memory formation under suggestive therapeutic pressure would be a substantive concern, and suggest targeted studies to assess that risk.