This review is reported according to the PRISMA statement (Moher et al., 2009). We performed electronic searches in Medline, Web of Science, and embase, from the respective database inception to November 18, 2020. Additionally, we searched the clinical trial registries ClinicalTrials.gov and EudraCT. The search was conducted using an algorithm connecting a selection of SPs and terms associated with neuropsychological testing or domains (shown in Supplemental Table S1) in an iterative manner. Given the broad variety of different available SPs, only those substances were included that exhibit a relevant degree of popularity and use prevalence in the population, such as the three most commonly used SPs, LSD, psilocybin and 2C-B (Evens et al., under review). Since we did not expect to find many relevant studies and because of the similarities in subjective experience and acute neuropsychological effects across substances discussed above, we decided to extend our search to all SPs across chemical sub-groups. References were retrieved through the electronic searches and by manual searches through the reference lists of review articles. All articles published in English, German, French, or Spanish were included. The PICOS (population, intervention, comparisons, outcomes, study type) selection criteria for our search are described in the supplementary methods section.

All search results were screened independently by two researchers (LAB, TGR), while a third (TM) provided input if it was not clear whether an article should be included or excluded. From the selected articles we recorded authors’ names, year of publication, duration and frequency of drug exposure, drug dosages, sample size, participant characteristics (number of female and male participants, mean age, age range), neuropsychological tests that were employed, and their results. In cases where data was missing for a study to be included in the meta-analysis, we contacted the authors. According to the content of the studies that were detected, the sample of studies was then divided into groups by substance, resulting in four groups: “LSD”, “Ayahuasca/DMT”, “Peyote/Mescaline”, and “Not specified”. Finally, the neuropsychological tests used in the studies were categorized into six domains: Memory, Executive Functioning, Attention, Visuospatial Abilities, Intelligence, and Other.

We used the Newcastle-Ottowa-Scale (NOS) (Wells et al., 2000) to estimate study quality and the risk of bias. The scale assigns a score for three parameters: “selection”, “comparability”, “outcome”, with a maximum total score of 9. We considered a study to be of high quality if it fulfilled both comparability criteria and reached a total score of seven or higher. LAB and TGR rated study quality independently, and subsequently formed a consensus on the rating of each parameter. Discrepancies were solved with input from TM.

Studies that reported results as mean scores with standard deviations were eligible for meta-analysis. For each neuropsychological test and subtest, a test of overall effect across means was conducted if at least three studies were available reporting mean and standard deviation for that test. In cases where different studies reported different outcome measures for the same test (for example, number of errors vs reaction time), the overall effect for the standardized mean difference was calculated instead. All calculations were performed with Review Manager 5.3 (The Cochrane Collaboration, 2014). In line with previous recommendations, we did not adjust p-values for multiple comparisons, due to the exploratory nature of our study (Rothman, 1990). We intended to perform four a-priori planned sensitivity analyses: 1) restriction to the studies with a rating of seven or higher on the NOS, 2) selective inclusion of studies employing matched control groups, 3) restriction to those studies examining the same SP, and 4) exclusion of the study with the greatest weight. In addition, heterogeneity in effect sizes was assessed using the I² statistic.

Excluding duplicates, our search identified 5,401 articles. Of these, 4,980 were rejected after title and abstract screening for not dealing with the effects of SPs on neuropsychological functioning, leaving 421 articles for full-text screening. After subsequent full-text screening, a total of 13 studies were left for inclusion in the systematic review (Cohen and Edwards, 1969; McGlothlin et al., 1969; Wright and Hogan, 1972; Culver and King, 1974; Matefy et al., 1979; Vardy and Kay, 1983; Grob et al., 1996; Doering-Silveira et al., 2005b; Halpern et al., 2005; Bouso et al., 2012, 2015; Barbosa et al., 2016; Kaasik and Kreegipuu, 2020). Five of these articles met the criteria for inclusion in the meta-analysis (Grob et al., 1996; Doering-Silveira et al., 2005b; Halpern et al., 2005; Bouso et al., 2012; Barbosa et al., 2016). Details of the different phases of the search are shown in Figure 1.

All included studies were cross-sectional studies, comparing cohorts of users of SPs (n = 539) to various groups of non-users. Five of the selected studies investigated the effects of repeated LSD use (n = 101) (Cohen and Edwards, 1969; McGlothlin et al., 1969; Wright and Hogan, 1972; Culver and King, 1974; Vardy and Kay, 1983), six explored the effects of ayahuasca (n = 343) (Grob et al., 1996; Doering-Silveira et al., 2005b; Bouso et al., 2012, 2015; Barbosa et al., 2016; Kaasik and Kreegipuu, 2020), one dealt with peyote users (n = 61) (Halpern et al., 2005), and one study did not specify which SPs had been used by the participants (n = 34) (Matefy et al., 1979). The participant demographics, study characteristics, and neuropsychological tests results for all included studies are shown in Table 1.

The median NOS score across all studies was 5, with one study receiving the lowest score of 3 (Kaasik and Kreegipuu, 2020) and one study receiving the highest score of 7 (McGlothlin et al., 1969) but not achieving a full comparability score, meaning no included study was rated as having high quality. The most common sources of bias were lack of an objective verification of SP exposure, with no studies providing this, and reduced comparability by not controlling for other substance use, with only three studies fulfilling this requirement (Culver and King, 1974; Halpern et al., 2005; Bouso et al., 2015). An overview of the complete ratings is given in Supplementary Table S2.

A complete overview of the measures on which a meta-analysis was performed, the number of included participants, the statistical method used, heterogeneity, and the effect estimates can be found in Tables 2–5. Forest plots for all performed analyses are shown in Supplementary Figures S1–4.

In the first identified period of research, LSD was by far the most intensively used SP, which is reflected in the neuropsychological studies predominantly focusing on LSD. As mentioned above, the majority of studies were of insufficient quality and did not adequately control for other, non-psychedelic substances. Arguably the first controlled study investigating neuropsychological consequences of SP use was conducted by Cohen and Edwards (1969), who compared 30 users of LSD (with a median of 70 LSD exposures, see Table 1) to 30 controls matched on gender, age, educational level, and socio-economic background and found that LSD users showed lower attentional and visuo-spatial performance (see Table 1 for details). However, the LSD group had not only taken LSD more often than the control group, but had also used more cannabis, amphetamines, barbiturates, heroin, and cocaine. Based on the results of this study, McGlothlin et al. (1969) compared 16 participants with a history of LSD exposure (partly in a therapeutic setting; median 75 LSD exposures) and compared these with 16 controls matched on age, gender, education, occupation, and the number of people who had received psychotherapy (without LSD). The authors report reduced performance in a categorical task for LSD users. Even though this study received the highest rating regarding study quality, it did not include sufficient matching in terms of substance use. Four participants of the control groups had previously used cannabis ten or more times, while in the LSD group eight had done so. Additionally, six participants in the LSD group had low to moderate use of opiates, sedatives and stimulants, but the history of use of those substances was not quantified (McGlothlin et al., 1969, p.2). In a similar fashion, Wright and Hogan (1972) compared 20 frequent recreational LSD users (mean number of LSD exposures = 29.3) with 20 controls, matched for gender, age, education, and intelligence, and found LSD users to perform better on one but worse on another subtest of the WAIS. It was reported that members of the control groups had not used any drugs, while of the 20 members of the LSD group, 19 reported cannabis use, 10 methamphetamine use, five opium use, and a few other substances (cocaine, various stimulant medications, various sedatives or opiates) had each been used by less than five participants. Another study (Culver and King, 1974) aimed to compare LSD users with non-users, recruiting 14 LSD users with additional cannabis use (and a median of 17 LSD experiences), 14 cannabis users, and 14 controls without a history of LSD or cannabis use. Additionally, they matched participants in all three groups according to verbal aptitude, mathematical ability, and personality profiles, and excluded participants for regular use of other substances. Furthermore, the cannabis group and the LSD group reported similar amounts of lifetime cannabis and alcohol use. In their investigation, they found that the LSD users performed worse than the cannabis users in tests of attentional performance and executive functioning (trials A and B of the Trail Making Test), indicating that this difference might be due to the use of LSD.

Apart from this apparent negative association of LSD use with neuropsychological functioning, LSD is the substance for which association with Hallucinogen Persisting Perception Disorder (HPPD) has most commonly been reported–even if this might reflect the overall high frequency of LSD use when compared to other SPs, and not a relative risk (Martinotti et al., 2018). Trying to investigate attentional performance of SP users with flashbacks, Matefy et al. (1979) compared 29 SP users with flashbacks, 25 SP users without flashbacks, and 23 controls without any substance use on a simple reaction-time task. All three groups were similar in terms of age, sex, hobbies, education, and their father’s education. SP users showed an increased reaction time across five measures. However, in an earlier publication, the authors report that the two SP groups were using cannabis in addition to SPs and had previously used sedatives, stimulants, cocaine, and heroin (Matefy et al., 1978). Standing out from these previous lines of research dealing with mostly healthy recreational users, Vardy and Kay (1983) compared neuropsychological performance in 29 patients who had been hospitalized because of psychotic symptoms following LSD use with 29 patients with schizophrenia without any history of LSD or other drug use. The authors report no differences in their neuropsychological tests. However, they acknowledge that the patients with a history of LSD use also report “more general drug experience”, without clarifying or quantifying history of drug use (Vardy and Kay, 1983, p. 2). Due to its focus on patient populations, this study is markedly different from the other studies discussed here, and its results cannot be generalized to healthy participants.

After this early period of LSD research, no further studies have been published dealing with the neuropsychological consequences of regular LSD use. Given the high overall polyvalent substance use of LSD users associated with the aforementioned methodological problems, studies from the 1990s started to investigate religious or ethnic groups ritually using specific SPs but having low use of other substances. One of these studies, and indeed the only investigation of a phenethylamine SP included in our review, Halpern et al. (2005) compared 61 religious users of peyote (with a median of 300 peyote experiences) with 36 former patients suffering from alcohol use disorder and 79 controls with minimal substance use. To control for other substance use, participants were only included if they reported lifetime use of cocaine, stimulants, opioids, sedatives, other SPs than peyote, or inhalants less than ten times, and less than 100 lifetime occasions of cannabis use. With these rigorous controls in place, the authors report no differences in any of the eight neuropsychological tests between the peyote group and the control group.

Investigating a different kind of religious SP use, several studies have studied consequences of regular ayahuasca use in members of syncretic Brazilian churches that regularly use ayahuasca as a sacrament in religious rituals. Among these, the church União do Vegetal (UDV) has been considered particularly useful for research (Grob et al., 1996; Doering-Silveira et al., 2005b; Barbosa et al., 2016), as UDV members are required to abstain from the use of any other substances, including alcohol and cannabis. Consequently, UDV members show very high lifetime use of ayahuasca and usually comparably low use of other substances. In a first study, 15 male members of the UDV (with at least 240 ayahuasca experiences, see Table 1) were compared with 15 male controls, who were closely matched to the ayahuasca group in terms of age, ethnicity, marital status, and level of education (Grob et al., 1996). The authors found that the ayahuasca users performed better on one measure of verbal memory. Matching for the use of other substances was not entirely successful: While the members of the ayahuasca group had been church members for at least 10 years and therefore abstinent from other substances for that time, eleven members reported moderate to severe alcohol use and five reported a history of cocaine and amphetamine use before their engagement with the church. Additionally, two members of the control group reported a current alcohol use disorder. Another study (Doering-Silveira et al., 2005b) investigated the differences in neuropsychological performance between 40 adolescent members of the UDV church (with at least 24 ayahuasca experiences) and 40 adolescent non-users matched to the ayahuasca group in terms of age, gender, race, and educational level, also reporting an increased performance for ayahuasca users in verbal memory. Still, a previous investigation of these two groups showed that they differed significantly in the prevalence of different psychoactive substance use, mainly alcohol, amphetamine, and solvents (Doering-Silveira et al., 2005a). Furthermore, Barbosa et al. (2016) investigated another 30 members of the UDV church (with a median of 150 ayahuasca exposures) and matched them to a control group without a history of ayahuasca use of 27 in terms of membership in a religious organization, age, and gender. In this study, the authors likewise report an increased verbal learning performance in ayahuasca users. Similarly to previous research with UDV members (Grob et al., 1996), the authors found that the members of the ayahuasca group had a higher lifetime exposure to other substances, in this case alcohol and cannabis. Additionally, the control group had used significantly more alcohol in the past month compared to the UDV group.

In a larger study with members of the Santo Daime ayahuasca church, which has less intense restrictions on the use of other substances, Bouso et al. (2012) further investigated neuropsychological consequences of ayahuasca use. They recruited ayahuasca users (with 260–1,440 experiences) from a community within the Amazon rain forest (n = 56) as well as users from an urban setting (n = 71) and compared these groups to controls from similar settings (n = 56 from a town close to the Amazon group, and n = 59 from the same city as the urban group), finding a higher level of executive functioning in ayahuasca users. However, in a previous study using the same groups, the authors reported that both ayahuasca groups had more past-month cannabis use and higher lifetime amphetamine and cocaine use than the control groups (Fábregas et al., 2010). Finally, the most recent study included in our review compared 30 ayahuasca users (with a median of 10 exposures) with 30 non-users in Estonia (Kaasik and Kreegipuu, 2020), with no significant differences in intelligence. Similarly to Bouso et al. (2012), they also found that their ayahuasca-using participants had used more cannabis than their control group. Across the studies conducted in religious settings, it becomes clear that even those subjects are often not free from a history of polysubstance use, and that even recruitment from churches with presently abstinent users does not always allow for successful control in terms of co-occurring substance use.

A single study, Bouso et al. (2015), explicitly required participants in both study and control groups to report lifetime cannabis use on less than 20 occasions and lifetime use of other substances on less than ten occasions. In the study, 22 ayahuasca users (with a mean of 123 ayahuasca exposures) outperformed 22 controls on tasks related to working memory (two-back task) and executive functioning (task-switching).

This association between ayahuasca use and improved executive control is partially supported by our meta-analysis, which showed that ayahuasca users consistently performed better in the inhibitory control section of the Stroop task. Taken together with the results of the well-controlled study by Bouso et al. (2015), these results may hint at a beneficial effect of ayahuasca use on executive functioning. This conclusion is surprising insofar as regular use of other substances such as alcohol, cocaine, or methamphetamine associates with impairments in executive functioning (Verdejo-García et al., 2006; Fernández-Serrano et al., 2009; Le Berre et al., 2010; Nestor et al., 2011). One explanation for this proposed effect of ayahuasca use could be related to its pharmacological mechanism. Tryptamines, such as DMT, show a higher binding affinity to the 5-HT_1aR than the 5-HT_2aR most SPs bind to (Fantegrossi et al., 2008; Winter 2009). Furthermore, tryptamines involve more 5-HT_1aR agonism than phenethylamines (Halberstadt and Geyer, 2011). The 5-HT_1aR has been shown to play a role in cognitive control in both humans (Langenecker et al., 2019) and animals (Baba et al., 2015). In addition, use of a 5-HT_1a agonist over the course of 6 weeks has been shown to increase executive functioning in patients with schizophrenia (Sumiyoshi et al., 2001). Regular administration of ayahuasca, an agonist at the 5-HT_1aR may induce similar improvements. This hypothesis could also explain why a similar pattern was not found in well-controlled studies with users of LSD (Culver and King, 1974) or peyote (Halpern et al., 2005), as both substances differ significantly from members of the tryptamine class in their receptor binding profiles (Ray, 2010). Another explanation for these results might be related to the involvement of specific brain structures. It has been reported that performance in the Stroop task was associated with activation of the anterior cingulate cortex (ACC) (Peterson et al., 1999; Ridderinkhof et al., 2004; Tolomeo et al., 2016) and Bouso et al. (2015) showed that use of ayahuasca was associated with higher cortical thickness in the ACC. We hypothesize that structural differences in specific brain regions between ayahuasca users and non-users might contribute to the improved performance in the Stroop task in users, however, this interpretation is highly preliminary and requires further research. Finally, since behaviors that improve neuronal plasticity are associated with improved cognitive performance (Greenwood and Parasuraman, 2010), and SPs have been shown to induce structural and functional plasticity (Marinova et al., 2017; Ly et al., 2018), it is not surprising that SP use in some domains might be associated with neuropsychological improvement.

Although these may be intriguing hypotheses, it bears reiterating that most research in this field is cross-sectional, not allowing any conclusions on causality. Surprisingly, even though psilocybin has been the most intensively investigated SP in humans in the past 25 years (Johnson and Griffiths, 2017), we could not find any studies assessing the neuropsychological consequences of its use. As psilocybin and its active metabolite psilocin show a similar binding affinity to the 5HT1a receptor (Rickli et al., 2016) and are structurally very similar to DMT (see Figure 2), the main psychoactive component of ayahuasca, which was investigated in several of our included studies, we speculate that their neuropsychological consequences could be of a similar nature. However, studies with well-matched control groups are necessary before any conclusions about the neuropsychological effects of psilocybin can be drawn. Specifically, there is a need for studies taking strong measures to control for potential confounding factors, especially with regard to substance use. Matching the control group and exposure group should be of highest priority when designing a study to establish consequences of repeated use. Since all modern studies included in our review have been conducted in ritual users of mescaline or ayahuasca, future research should take care to include users of psilocybin, as this is one of the most frequently used SPs in recreational settings (Krebs and Johansen, 2013) and has been most extensively investigated in clinical studies (Bogenschutz and Ross, 2018). Notably, the relevance of our findings regarding SP-assisted therapy remains limited, as users in our studies mostly used SPs repeatedly in recreational or traditional context, without any psychotherapeutic support and without knowledge on the exact concentration and purity of substances. Clinical use of SPs involves pharmacologically pure substances of a defined dosage, whereas in recreational use, the consumed substance is often not reliably defined (Hirschfeld et al., 2021) and even in traditional use, dosage and composition might often underlie strong variations (Gaujac et al., 2013). Nevertheless, many of the participants in the identified studies reported excessive lifetime use of SPs, by far exceeding SP exposition of patients in clinical studies of SP-assisted therapy, where substances are administered only very few times, never reaching the lifetime use of participants in the reported studies. In conclusion, as heavy use was not associated with decreases in neuropsychological outcome, it appears unlikely that negative effects would be found when SPs are administered only rarely, as it is the case for SP-assisted psychotherapy.

As mentioned above, a major limitation of this review is the difficulty of controlling for the use of other psychoactive substances evident in the included studies. Implementing this control is challenging, as the majority of SP users report use of additional psychoactive substances, especially cannabis (Pisano et al., 2017). In trying to restrict the studies to participants taking SPs exclusively, some studies recruited their samples in groups of a specific religious background who use SPs as part of their religious practice, leading to a very narrow selection of participant demographics. In fact, nearly all studies in this field conducted after 1990 are limited to the use of ayahuasca or peyote in ritualized/religious settings.

Although substances belonging to the group of SPs may overlap regarding phenomenological aspects of associated psychedelic experiences, generalization of the findings of possible beneficial effects of ayahuasca to the whole group of SPs (and even to all tryptamine psychedelics) remains problematic. This is starkly illustrated by the fact that one well-controlled study involving LSD users (Culver and King, 1974) reported diminished performance in tasks dealing with executive functioning and attention, and Halpern et al. (2005) reported no impairments or improvements in peyote users. Furthermore, the analyzed studies in our meta-analysis of Stroop task performance (Grob et al., 1996; Doering-Silveira et al., 2005b; Bouso et al., 2012) did not entirely succeed in controlling for other substance use. Therefore, it cannot be ruled out that the better performance in the ayahuasca-using group might also reflect effects of other substances, or other non-substance related differences between groups.