No evidence that LSD microdosing affects recall or the balance between distracter resistance and updating

Interest in LSD and other classic psychedelics has re‑emerged, yet their cognitive effects—especially at very low ‘‘microdose’’ levels that do not produce overt changes in consciousness—remain incompletely characterised. Working memory, the short‑term retention and manipulation of information needed for goal‑directed behaviour, is a key cognitive domain that must balance stability (resisting distraction) and flexibility (updating representations). Previous work implicates both dopaminergic and serotonergic (notably 5‑HT2A) signalling in these processes, and computational models predict that pharmacological modulation can produce opposing effects on distracter resistance versus updating. Despite widespread anecdotal reports that microdosing LSD improves cognition, few placebo‑controlled laboratory studies have examined these hypotheses and tasks such as the n‑back conflate stability and flexibility, limiting inference about dissociable effects. Fallon and colleagues set out to test whether very low doses of LSD (5 µg, 10 µg and 20 µg) alter working memory performance in older adults and whether any effects are differentially expressed for ignoring distractors versus updating working memory. They used a randomised, double‑blind, placebo‑controlled design and a modified delay‑match‑to‑sample task that explicitly separates trials requiring maintenance, ignoring novel stimuli, or updating with novel stimuli, enabling direct assessment of stability versus flexibility in working memory.

The study was embedded in a larger safety/tolerability trial and used a randomised, double‑blind, placebo‑controlled design. Forty‑eight native English‑speaking older adults (21 female, 27 male) aged between 55 and 75 years were randomised equally into four arms (12 per arm) to receive placebo or one of three LSD microdoses (5 µg, 10 µg, 20 µg). The extracted text does not clearly report the sample mean age (the reported value appears corrupted). Recruitment occurred at the Early Phase Clinical Unit, Northwich Park, UK. Participants had not used LSD in the preceding five years and additional contraceptive, health and medication restrictions applied. Key inclusion and exclusion criteria are reported; selected points include: - Inclusion: native English speakers, age 55–75, no LSD use in last five years; females were post‑menopausal and males had to agree contraception measures as specified. - Exclusion: history of major medical or psychiatric disorders; elevated blood pressure above screening thresholds; abnormal laboratory tests or positive viral serology; recent substance use or smoking history; recent use of serotonin‑modifying medications; blood donation restrictions; lifetime non‑drug‑induced psychosis; inability to use a computer for cognitive testing. The principal cognitive assay was a modified delay‑match‑to‑sample (DMTS) task designed to dissociate maintenance, ignoring and updating. Each trial began with encoding two stimuli (2000 ms), then a delay (4000 ms) during which participants either saw a blank fixation (maintain), novel stimuli that should be ignored (ignore), or novel stimuli that should be updated into working memory as the new targets (update). Participants were implicitly cued to treat stimuli with the letter ‘‘T’’ as targets; they were not explicitly told to ignore or update. After a second delay a probe item was presented and participants indicated whether it matched the relevant target(s). Stimuli were abstract ‘‘spirographs’’. Each participant completed 72 trials (24 per condition), with half match and half non‑match trials; non‑match probes drew equally from novel and previously presented but now non‑target stimuli. Analyses used mixed generalised linear models implemented in Matlab (fitlme/fitglme). Trial‑level accuracy (correct/incorrect) was modelled with a binomial distribution and logit link (a generalised linear mixed model). Response latencies for correct trials were analysed with a linear mixed model assuming a normal distribution. Condition (ignore, maintain, update) and drug condition were entered as fixed effects; participant was included as a random intercept (with participants split according to drug condition in the random effect structure as described in the extracted text).

Performance varied reliably by task condition but not by drug. A generalised mixed model found a significant main effect of condition on accuracy (F(2,3404) = 9.51, p < .0001). Participants were less accurate on ignore trials compared with update trials (F(1,2273) = 12.97, p = .0003) and less accurate on ignore trials compared with maintain trials (F(1,2267) = 14.64, p = .0001). There was no evidence for a difference between update and maintain trials (F < 1). Crucially, there was no evidence that LSD microdose altered accuracy: the main effect of drug dose was nonsignificant (F(3,3404) = 1.11, p = .34), and there was no significant interaction between condition and drug dose (F < 1). Reaction time analyses on correct trials showed a similar pattern: a significant main effect of condition (F(2,2748) = 14.91, p < .0001) but no effect of drug dose and no drug × condition interaction (reported as Fs < 1). In short, the task revealed expected differences between ignoring, maintaining and updating demands, but no detectable influence of 5 µg, 10 µg or 20 µg LSD on either accuracy or response latency in this sample.

Fallon interprets the findings as showing no evidence that LSD microdoses, at the tested levels, affect overall working memory recall or the balance between stability (distracter resistance) and flexibility (updating) of mental representations. The absence of drug effects aligns with prior studies that failed to detect mnemonic changes with LSD microdoses, although the literature is limited and mixed for low doses more generally. The authors note that both serotonergic (5‑HT2A) and dopaminergic pathways have theoretical and empirical links to working memory and to the stability–flexibility trade‑off, so pharmacological modulation could reasonably have been expected to influence task performance. They argue that the modified DMTS paradigm used here is sensitive to dissociable effects on ignoring and updating—citing prior pharmacological work showing similar designs can detect drug‑specific effects—so task insensitivity is an unlikely explanation for the null drug findings. Limitations acknowledged by the study team include the modest sample size, which reduces statistical power and precludes strong inferences from the null results. Fallon and colleagues point out that other outcomes in the broader trial did show drug‑related effects (for example, on time perception), suggesting that LSD microdoses may have larger effects on some domains than others and that the impact on working memory, if present, may be smaller. The authors also highlight the potential for baseline‑dependent drug effects—where individual traits such as impulsivity, genetic background or baseline working memory capacity moderate response to dopaminergic or serotonergic agents—and recommend larger studies capable of characterising individual differences in drug response to provide a more definitive answer. In conclusion, within the constraints of this sample and task, the study found no evidence that very low doses of LSD altered working memory accuracy or reaction time, nor that LSD differentially affected ignoring versus updating; the authors call for larger, more powered investigations that can account for baseline dependence and individual variability.