Neuropsychological Functioning in Users of Serotonergic Psychedelics - A Systematic Review and Meta-Analysis

Basedow and colleagues situate this review in the context of a renewed scientific interest in serotonergic psychedelics (SPs) such as LSD, psilocybin, and DMT. They note that SPs are a pharmacologically heterogeneous group that share agonism at the 5-HT2a receptor but differ in chemical class (tryptamines, ergolines, phenethylamines), binding affinities, receptor selectivity and intracellular signalling. Although acute psychedelic states produce transient neuropsychological impairment, SPs have also been reported to produce persisting psychological and trait-like changes, and rare prolonged adverse reactions such as hallucinogen persisting perception disorder (HPPD). Despite this, the authors state that neuropsychological consequences of repeated SP use remain underexplored and previous systematic reviews were limited by the methodological quality of included studies. This paper therefore aims to determine whether repeated SP use is associated with a characteristic profile of persisting neuropsychological effects. To that end, the researchers conducted a systematic review and an exploratory meta-analysis of studies reporting neuropsychological test performance in SP users versus non-users, with the intention of assessing across multiple cognitive domains whether consistent deficits or enhancements emerge following repeated exposure.

The review was conducted according to PRISMA principles. Electronic searches of Medline, Web of Science and embase were performed from database inception to 18 November 2020; the clinical trial registries ClinicalTrials.gov and EudraCT were also searched. Searches combined names of common SPs and terms for neuropsychological tests/domains; articles in English, German, French or Spanish were eligible. Further details on PICOS criteria are reported in the supplementary methods (not reproduced here). Screening and extraction were performed independently by two researchers (LAB, TGR) with a third (TM) adjudicating unclear cases. Extracted items included publication details, exposure duration and frequency, dosages when reported, sample demographics, neuropsychological tests used and outcomes. Detected studies were grouped by substance into four categories: LSD; Ayahuasca/DMT; Peyote/Mescaline; and Not specified. Neuropsychological measures were assigned to six domains: Memory, Executive Functioning, Attention, Visuospatial Abilities, Intelligence, and Other. Study quality and risk of bias were assessed using the Newcastle–Ottawa Scale (NOS), which rates selection, comparability and outcome up to a maximum of 9 points; studies scoring 7+ with both comparability criteria met were considered high quality. For the meta-analysis, studies reporting means and standard deviations were eligible. A minimum of three studies reporting a given test was required to compute an overall effect for that test; when studies reported different outcome metrics for the same test the standardised mean difference was calculated. Analyses used Review Manager 5.3. Given the exploratory approach, p-values were not adjusted for multiple comparisons. Four a priori sensitivity analyses were planned: restriction to higher-quality studies (NOS ≥7), inclusion of matched-control studies only, restriction to the same SP, and exclusion of the study with greatest weight. Heterogeneity was to be assessed using the I2 statistic. The authors report that all included studies were cross-sectional comparisons of SP users versus various non-user groups.

The searches identified 5,401 records after de-duplication; 421 underwent full-text screening and 13 studies (total N = 539 participants) met inclusion criteria for the systematic review. Five of these studies contributed data to meta-analyses. Study types were all cross-sectional; substance-specific study counts were: ayahuasca 6 studies (n = 343), LSD 5 studies (n = 101), peyote 1 study (n = 61) and 1 study with unspecified SPs (n = 34). The median NOS score across studies was 5 (range 3–7); no study met the review’s criteria for high quality. The most common biases were lack of objective verification of SP exposure and inadequate control for other psychoactive substance use, with only three studies controlling sufficiently for co-occurring substance use. Memory: Working-memory assessments included WAIS digit span (four studies) and a two-back task (one study). Across these, samples comprised 157 SP users (34 LSD, 62 ayahuasca, 61 peyote) and 175 controls. The two-back task showed better performance in ayahuasca users (higher hits, fewer misses, faster reaction times), but there were no group differences in false alarms or correct rejections. Verbal episodic memory (UCLA AVLT/CVLT and visual recall tasks) comparisons yielded inconsistent single-trial advantages for ayahuasca users on particular trials in some studies, but meta-analysis of delayed recall across four studies (305 participants: SP n = 145, controls n = 160) found no statistically significant difference (Z = 0.08; p = 0.94). Other verbal learning comparisons (trial 5, short recall after interference, recognition) were likewise non-significant (all p ≥ 0.68). Executive functioning: A variety of tasks were used (letter-number sequencing, WAIS subtests, Stroop, WCST, set-shifting). Individual study results were mixed: some older LSD studies reported worse performance on Trail Making Test B (TMT-B), whereas several ayahuasca studies reported better performance on various executive tasks. Meta-analysis of TMT-B across three studies (276 participants: SP n = 131, controls n = 145) showed no overall difference (Z = 0.01; p = 0.99). For the Stroop task, four studies (including two independent ayahuasca samples treated separately) were pooled using standardised mean differences because of measurement differences; across 514 participants (SP n = 257, controls n = 257) the incongruent list subtest showed better performance in SP users (Z = 2.14; p = 0.03), whereas word and colour lists were non-significant (p = 0.85 and p = 0.24). A planned sensitivity analysis restricting to the same SP increased the Stroop incongruent effect (p = 0.004). Attention: TMT-A was pooled across three studies (276 participants) with no significant group difference (Z = 0.03; p = 0.98). Some single studies reported poorer TMT-A performance in LSD users and faster simple reaction times in one SP group, but most attention measures were null. Visuospatial abilities and intelligence: Meta-analysis of the Rey–Osterrieth Complex Figure copy and delayed recall across three studies (276 participants) were non-significant (copy: Z = 1.13; p = 0.26; delayed recall: Z = 1.46; p = 0.14). Intelligence measures (Shipley, WAIS, Raven) showed no group differences across six studies. Other assorted subtests produced a few isolated significant differences in older LSD studies, but results were inconsistent across investigations. Overall quantitative assessment was limited: only five studies contributed to meta-analyses for specific tests, and many individual studies differed in outcome metrics or procedures. Reported heterogeneity assessments were planned using I2, but specific I2 values are not clearly reported in the extracted text. Importantly, most studies did not fully control for co-occurring substance use, which the authors note weakens causal inference.

Basedow and colleagues interpret the body of evidence as lacking a uniform neuropsychological profile associated with repeated SP use; instead, they report substance-specific patterns with conflicting results across eras of research. Studies from 1969–1983 predominantly investigated LSD in recreational users who frequently used other psychoactive substances, and several of these older studies associated LSD with poorer attention and executive functioning. By contrast, studies from 1996–2020 largely examined ritualised ayahuasca or peyote users; several ayahuasca studies reported enhanced performance on some executive measures, whereas the single well-controlled peyote study found no differences. The authors discuss possible mechanistic explanations for the observed ayahuasca–executive-function association. Tryptamine SPs such as DMT show greater 5-HT1a receptor agonism than phenethylamines, and 5-HT1a activity has been implicated in cognitive control; structural differences such as increased anterior cingulate cortex thickness reported in some ayahuasca users might also relate to Stroop performance. In addition, SP-induced neuroplasticity could, in theory, promote cognitive improvements. Nevertheless, the investigators emphasise that most included studies are cross-sectional and therefore cannot establish causality. They also highlight a surprising absence of studies on psilocybin despite its prominence in recent clinical research. Key limitations acknowledged by the authors include pervasive polysubstance use among participants and imperfect matching or control for other psychoactive substances, small sample sizes, heterogeneity of neuropsychological measures and outcomes, and reliance on naturalistic samples with variable substance composition and dosing. Religious or ritualised user groups reduce some confounding but may introduce selection biases and narrow demographic generalisability. Given these constraints, the authors caution that the meta-analytic finding of better Stroop incongruent performance in SP users—primarily driven by ayahuasca samples—should be considered preliminary. They recommend future studies with robust control for confounders (especially other substance use), inclusion of psilocybin and other under-studied SPs, and designs capable of addressing causality such as longitudinal or well-matched cohort studies.

The authors conclude that reliable data on the neuropsychological consequences of repeated SP use are scarce and methodologically limited. No studies assessing psilocybin use were identified. Although some well-controlled studies associated LSD with reduced task-switching performance and ayahuasca with improved inhibitory control, peyote showed no relationship with neuropsychological outcomes in the available data. The difficulty of adequately controlling for co-occurring psychoactive substance use limits inference, and future research should clarify whether observed differences reflect distinct pharmacological actions across SPs.