The induction of synaesthesia with chemical agents: a systematic review

The authors conducted an English-language literature search using PubMed, PsychNet and PsycINFO with search terms including synaesthesia/synesthesia and a range of drug-related keywords (e.g., psychedelic, LSD, psilocybin, mescaline, MDMA, ketamine, cannabis). They also followed reference cascades from retrieved articles. Initially a meta-analysis was planned, but the small number of direct experimental attempts (four) precluded quantitative pooling. Studies were categorised into four design types for synthesis: direct experimental studies that explicitly attempted to induce synaesthesia, indirect experimental studies that included synaesthesia-related questionnaire items within broader assessments (some including placebo controls), case reports and anecdotal observations, and survey studies of substance users or of synaesthetes reporting drug effects. Inclusion therefore spanned formal experiments, controlled trials with ancillary synaesthesia measures, individual case descriptions, and prevalence surveys. The authors summarised and compared findings across these designs rather than applying a single analytic model. The extracted text does not report specific dates of the search or formal risk-of-bias assessment methods beyond describing methodological limitations qualitatively.

The search identified 35 studies overall, which the authors organised by design. Four direct experimental studies explicitly attempted pharmacological induction of synaesthesia. In one small study, five participants who had undergone tone–colour associative training consumed peyote (estimated mescaline content between ~0.15 and 1.2 g); four reported colourful visual imagery but not the trained tone–colour matches, instead reporting other forms such as haptic–visual, kinaesthetic–visual and algesic–colour experiences. A second experiment administered mescaline (0.3–0.5 g on separate occasions) to two congenital synaesthetes and non-synaesthete controls; across four participants novel synaesthesias were reported (auditory–visual in all four, kinaesthetic–visual in three, tactile–visual in two, and several other types), and one congenital synaesthete reported enhancement of their usual synaesthesias. A third blinded experiment compared mescaline (5 mg/kg), psilocybin (150 mcg/kg) and LSD (reported as 1 mcg/kg) in 18 participants exposed to auditory tones; LSD and mescaline significantly increased colours and auditorily driven visual effects relative to baseline, whereas psilocybin produced a non-significant increase; fewer than 50% of participants experienced auditorily-induced synaesthesia under these psychedelics. A fourth, larger series of informal experiments and interviews (214 participants in 204 sessions) was reported to have elicited auditory–colour and auditory–gustatory synaesthesias with LSD, but methodological and dosage details were sparse. Nine indirect experimental studies were identified in which psychedelic drugs were administered under controlled conditions and participants answered phenomenological questionnaires. These studies—some with placebo controls—reported variable prevalence of drug-induced auditory–visual synaesthesia: pooled figures cited include MDMA 10%, ketamine 27% and psilocybin 37%. Psilocybin showed a dose–response relationship in several datasets: positive questionnaire responses for auditory–visual synaesthesia ranged from 0% at placebo and very low doses (45 mcg/kg) up to about 50% at 315 mcg/kg, with earlier reports of under 50% at 150 mcg/kg. Individual differences predicted susceptibility: higher absorption (a trait reflecting immersive imaginative engagement) was associated with greater likelihood of reporting drug-induced synaesthesia. Seventeen case reports were reviewed. Many simply documented the form of induced synaesthesia, most commonly auditory–visual types triggered by music or voices following substances such as mescaline, LSD, cannabis or ayahuasca. Some case reports provided richer phenomenology: variability was noted in whether visual concurrents appeared as internal mental images or as spatially localised percepts—paralleling the associator/projector distinction among congenital synaesthetes. Mescaline was described as producing complex multi-sensory fusions in some cases, and one report described consistent grapheme–colour experiences following melatonin that was supported by a behavioural texture-segregation measure. A few case reports described modulation of congenital synaesthesia by drugs: fluoxetine and bupropion were reported to inhibit an unspecified congenital synaesthesia in one case, suggesting serotonergic involvement. Survey studies provided prevalence estimates from broader user samples. One survey reported that psychedelic substances in general induced synaesthesia in 50% of users (62 respondents described 23 different types, with auditory–visual representing 52% of those reports). A cannabis user survey found 56% reported music–colour synaesthesia. Lifetime prevalence from surveys tended to be higher than rates observed in individual experimental trials. In a survey of 1,279 verified grapheme–colour synaesthetes, alcohol and caffeine were each reported to enhance synaesthesia in 9% of respondents and reduce it in 6% (alcohol) and 3% (caffeine); among six synaesthetes who reported LSD use, two (33%) described enhancement. Across the dataset, the most commonly reported induced form was auditory–visual synaesthesia (19 reports; 23%), auditory stimuli were the most frequent inducers (39 reports; 47%), and visual experiences the most frequent concurrents (43 reports; 52%). However, these proportions are biased by the frequent exclusive measurement of auditory–visual phenomena in indirect studies. The authors note that while many drug-induced forms resemble those seen in congenital synaesthesia (e.g., grapheme–colour, auditory–colour), chemically-induced reports also include unusual inducer–concurrent pairings not typically documented in congenital cases (for example, visual–thermal).

Luke and colleagues conclude that there is convergent evidence across multiple methodologies that various chemical agents—most notably serotonergic agonists such as LSD, mescaline, psilocybin and ayahuasca—can produce synaesthesia-like experiences. They emphasise that this convergence spans experimental reports, controlled indirect studies, case reports and surveys, and that some data also indicate similar modulatory effects on congenital synaesthesia. At the same time, the authors caution that the literature has substantial methodological limitations that preclude firm conclusions about whether chemically-induced experiences constitute genuine synaesthesia. Key shortcomings are the relative scarcity of direct experimental work, frequent absence of placebo controls and double-blinding, reliance on self-report rather than behavioural verification of automaticity or consistency, incomplete reporting of doses and time courses, and failure to account for individual differences such as absorption. Increased suggestibility associated with psychedelics raises particular concern about demand characteristics inflating reports. The review highlights a strong signal implicating serotonin—particularly 5-HT2A agonism—in induced synaesthesia, but also notes that non-serotonergic substances (e.g., ketamine, Salvia divinorum, cannabis, nitrous oxide) have been reported to elicit synaesthesia-like phenomena, suggesting multiple pharmacological routes to altered cross-modal experiences. The authors discuss how induced synaesthesias compare with congenital forms: induced experiences meet the criterion of conscious accessibility but, insofar as the available evidence shows, do not reliably meet criteria for automaticity, consistency and specificity; only one study reported a behavioural consistency measure (melatonin-induced grapheme–colour), and no study has rigorously validated induced synaesthesias against established behavioural tests. For future work the authors recommend placebo-controlled, double-blind experiments that directly compare chemical classes, incorporate objective measures of automaticity and consistency rather than sole reliance on self-report, include comprehensive phenomenological inventories, and explore neural substrates (for example with transcranial magnetic stimulation) to determine whether induced and congenital synaesthesias depend on similar cortical networks. They also suggest considering the possibility that some synaesthetic criteria (such as long-term consistency) may require consolidation time and therefore might be difficult to demonstrate immediately following acute drug effects. Overall, the authors regard the existing findings as sufficiently suggestive to warrant further rigorous investigation, while stopping short of asserting that chemical agents have been shown to induce bona fide synaesthesia.