Dose-related effects of salvinorin A in humans: dissociative, hallucinogenic, and memory effects

Salvinorin A is the principal psychoactive constituent of Salvia divinorum and is pharmacologically distinct from classic serotonergic hallucinogens: it is a non‑nitrogenous, selective kappa opioid receptor agonist with negligible activity at 5‑HT2A. Nonhuman studies have characterised its pharmacology and behavioural effects, and survey-based human reports describe intense, often idiosyncratic subjective effects, but controlled human laboratory data remain limited and sometimes inconsistent. Prior controlled work from this group and others suggested dose‑related subjective effects after inhalation, but questions remain about the phenomenology, dissociative and cognitive consequences, and the persistence of effects. Maclean and colleagues designed a double‑blind, placebo‑controlled human laboratory study to extend preliminary observations and to characterise dose‑related effects of inhaled salvinorin A across a range of outcomes. Their aims were to document time course and dose–response relationships for subjective, observer and physiological measures, to test effects on recall and recognition memory, to capture qualitative aspects of experience via open‑ended narratives, and to assess persisting effects approximately 1 month after exposure.

The study used a double‑blind, placebo‑controlled, within‑subject design in eight healthy adults with histories of hallucinogen use (mean age 26.8 years, range 21–35; three female). Participants reported prior lifetime classic hallucinogen use and prior Salvia divinorum use; they completed medical and psychiatric screening and agreed to refrain from illicit drug use during participation. Institutional review board approval and informed consent were obtained. Each participant completed up to 20 sessions across several weeks: 16 ascending doses of salvinorin A (0.375–21 μg/kg, adjusted for bodyweight) and four placebo sessions, with consecutive sessions separated by at least 1 day. Placebo sessions were inserted at random positions among each block of five sessions; participants were told each session could be salvinorin A or placebo but were not informed of the ascending schedule. Administration used a vapourisation/inhalation procedure: participants inhaled slowly for 40 s while salvinorin A was heated, then exhaled on cue. During sessions participants wore eyeshades, were seated semi‑upright and listened to relaxing instrumental music. Outcome measures collected at baseline and every 2 min for 60 min after administration included blood pressure and heart rate, verbally cued participant ratings of drug strength, and blinded monitor ratings of multiple dimensions (overall drug effect, unresponsiveness, anxiety/fear, distance from usual reality, paranoia, motor activity, visual effects, joy/peace, psychological and physical distress). Resting and kinetic tremor were rated before dosing and at 15 and 30 min. An auditory word recall and recognition task assessed memory: three target words were presented at 2, 4 and 6 min post‑dose and recall and recognition were tested at 15 min; outcomes were number recalled (0–3), recognition accuracy (d') and response bias (C). End‑of‑session measures included retrospective overall drug strength, liking/disliking, good/bad effects, and several validated questionnaires: ARCI, Hallucinogen Rating Scale (HRS), APZ, Mysticism Scale, States of Consciousness Questionnaire (SOCQ), Perception Scale, STAI (state anxiety) and a Quantitative Pharmacological Class Questionnaire (QPCQ) assessing similarity to 15 drug classes. Participants also provided open‑ended narratives. A follow‑up about 1 month later included BDI, trait/state STAI, POMS, BSI, the Persisting Effects Questionnaire and structured community observer ratings. Analyses used repeated measures regression (mixed models, AR(1) covariance) with dose as a within‑subjects factor across 16 dose levels including placebo; Dunnett's tests compared each dose to placebo (p<0.05). For time‑series measures analysis focussed on peak ratings within session. Linear regression was used to model dose effects for the word recall task and QPCQ; paired t tests assessed change from baseline to follow‑up. Some data limitations were reported: two male participants declined the highest planned dose (21 μg/kg), so analyses used a maximum dose of 19.5 μg/kg and complete data for 15 salvinorin A levels. In sessions where participants were unresponsive, a maximal drug‑strength rating (10) was imputed for participant‑rated drug strength. One male participant's ARCI and word recall data were missing, leaving N=7 for those analyses, and a small number of single‑session word‑recall values were missing for two participants.

Participant‑rated drug strength rose rapidly, peaking at the first post‑dose time point (2 min) and then decreasing toward baseline. Peak drug strength showed orderly dose‑related increases across the tested range: participant peak ratings F(15,105)=7.92, p<0.001; monitor peak ratings F(15,105)=3.38, p<0.001. End‑of‑session retrospective ratings showed a similar dose effect F(15,105)=10.8, p<0.001. Four participants provided the maximal rating of 10 and five participants were unresponsive at one or more time points (scores imputed as 10) for at least one dose. Two participants exhibited pronounced dissociative behavioural responses at doses above 15.0 μg/kg. Physiological measures showed no significant dose effects on blood pressure or heart rate (p>0.10). Tremor ratings were essentially absent; a single male had slight resting and kinetic tremor ratings at specific doses. Blinded monitor peak ratings revealed dose‑related increases in distance from usual daily reality (F(15,105)=2.92, p=0.001) and unresponsiveness (F(15,105)=1.81, p=0.043). Motor activity and psychological distress also showed significant dose effects (motor activity F(15,105)=1.81, p=0.043; psychological distress F(15,105)=2.44, p=0.004). Other monitored dimensions (joy/peace, anxiety/fear, paranoia, visual effects, physical distress) remained generally low and non‑significant. End‑of‑session liking and reports of good effects increased with dose (liking F(15,105)=3.65, p<0.001; good effects showed a trend F(15,105)=1.74, p=0.054). Mean disliking and bad‑effects ratings were low and unaffected by dose. Questionnaires assessing classic hallucinogen‑type and mystical effects (HRS, APZ, Mysticism Scale, SOCQ) showed significant dose effects, with total and most subscale scores elevated at moderate and high doses. The Perception Scale—designed to capture kappa opioid‑type effects—showed dose effects for total score and subscales including detachment, with detachment significantly above placebo for all doses ≥4.5 μg/kg. ARCI scales showed a dose‑related increase only on the Benzedrine Group subscale, but without a clear dose‑ordered pattern. State anxiety (STAI) did not change significantly. On the QPCQ, classic hallucinogen similarity ratings increased with dose (β=0.34, R2=0.11, p<0.001) but mean similarity remained low (<30/100) and highly variable across participants; at the two highest doses only two participants rated similarity >75/100 while five rated it <25/100. Memory was impaired in a dose‑dependent manner. Dose correlated negatively with number of words recalled (r=-0.49, p<0.001) and recognition accuracy d' (r=-0.42, p<0.001); recognition response bias C showed a small positive correlation with dose (r=0.24, p=0.012), indicating a more conservative response tendency at higher doses. Stepwise regression controlling for baseline performance found that dose explained significant additional variance in recall (β=-0.50, ΔR2=0.25, p<0.001) and d' (β=-0.43, ΔR2=0.18, p<0.001). Qualitative open‑ended narratives revealed five consistent themes: disruptions in vestibular/interoceptive signalling (e.g., sensations of being pulled, spinning, stretching), encounters or communication with entities, revisiting childhood memories (specific and scene‑like), cartoon‑like imagery and auditory experiences often linked to childhood, and recurring content across sessions. The extracted text includes representative brief verbatim excerpts illustrating these themes. At approximately 1‑month follow‑up there were no significant changes from screening on depressive symptoms (BDI), anxiety (STAI), affective mood state (POMS subscales reported), psychiatric symptoms (BSI) or visual disturbances (p values >0.20). One participant showed a >2 SD increase on the BSI but staff concluded this was unrelated to study participation. Participants rated sessions as personally meaningful (mean 4.9 on the provided scale) and spiritually significant (mean 3.6), and reported modest positive changes in well‑being and attitudes; four participants reported specific positive changes attributed to the sessions. Community observer ratings likewise indicated slight favourable changes in behaviour and mood.

Maclean and colleagues interpret their findings as demonstrating reliable, rapid‑onset and dose‑related subjective, observer and objective cognitive effects of inhaled salvinorin A in hallucinogen‑experienced adults. The time course was brief, peaking at about 2 min and dissipating rapidly. Dissociative phenomena were pronounced at moderate and high doses, evidenced by increases in monitor ratings of distance from daily reality and unresponsiveness, participant reports of detachment and altered sense of time/space, and dose‑related impairments in recall and recognition memory. Despite these acute effects, there were no systematic dose‑related changes in blood pressure, heart rate or tremor. The investigators note partial phenomenological overlap with classic serotonergic hallucinogens on measures such as the HRS and Mysticism Scale at higher doses, but emphasise that mean similarity ratings remained low and highly variable; participants predominantly described the effects as unique. Qualitative analysis highlighted themes—vestibular/interoceptive disruption, entity contact, childhood recollection and cartoon‑like imagery—that standard hallucinogen scales may not capture well. Accordingly, the authors suggest that existing instruments have limited utility for characterising kappa opioid‑mediated phenomenology and recommend development or validation of more diverse quantitative and qualitative measures as well as objective behavioural tasks. Regarding safety and abuse liability, liking and 'good effect' ratings increased with dose while disliking and state anxiety remained low, indicating some potential for reinforcing effects; the authors suggest future work should assess reinforcement directly (for example via self‑administration paradigms) and examine affective changes in more detail. They caution that intense acute dissociative reactions observed—including brief unresponsiveness and erratic movements in two participants at higher doses—highlight risks when salvinorin A is used in uncontrolled settings and support psychiatric screening and in‑session monitoring in human research. The authors acknowledge key limitations: a small, homogeneous sample of experienced hallucinogen users, missing data for some measures, and absence of a separate control group. Nonetheless, follow‑up assessments showed no evidence of persistent adverse psychological effects and several participants reported positive personal meaning. The inhalation method used here appeared more reliable than sublingual or other procedures reported in prior work, and the authors advise caution when comparing doses across administration techniques. They propose future studies including neuroimaging to map neural correlates of the reported phenomena, further refinement of outcome measures to capture kappa opioid‑specific effects, and studies across more diverse participant samples.