Investigating the Mechanisms of Hallucinogen-Induced Visions Using 3,4-Methylenedioxyamphetamine (MDA): A Randomized Controlled Trial in Humans

Serotonergic hallucinogens such as LSD, mescaline and psilocybin produce a wide range of visual phenomena, from altered form and depth perception to vivid pseudo-hallucinations that may occur with the eyes closed (closed-eye visions, CEVs) or open (open-eye visions, OEVs). The mechanisms underlying these drug-induced visual changes remain poorly understood; prevailing explanations implicate (1) reduced sensory or perceptual fidelity, (2) abnormally increased cortical excitation, and/or (3) altered cognitive or top-down influences on perception. Previous work has linked visual hallucinations in other conditions to low-level visual deficits (for example, impaired contour detection), to abnormal excitation as in migraine or epilepsy, and to altered use of top-down cues for recognition. Baggott and colleagues set out to probe these putative mechanisms by administering 3,4-methylenedioxyamphetamine (MDA) to healthy, drug-experienced volunteers in a controlled laboratory experiment. The study tested whether MDA would produce self-reported hallucinogen effects (including CEVs and mystical-type experiences) and whether those visual reports would be associated with changes on perceptual tasks designed to index reduced sensory fidelity (contour integration and object recognition) and altered cortical excitability (tilt illusion). The authors therefore aimed to relate subjective visual phenomena to objective measures of perceptual organisation and early visual cortical processing.

The study used a double-blind, placebo-controlled within-subjects crossover design conducted at a clinical research centre. Twelve healthy participants with prior experience of MDA, MDMA, or other hallucinogens completed two consecutive dosing sessions (MDA and placebo, order counterbalanced). Safety screening included medical history, EKG and blood work; pregnancy and toxicology testing were performed before dosing. Racemic MDA was synthesised for the study and administered orally at 98 mg per 70 kg body weight; lactose capsules served as placebo. Participants stayed for a single three-evening admission and returned two weeks later for follow-up checks. Timed measures included physiological monitoring, blood draws for pharmacokinetics, repeated self-report instruments and three computerised visual tasks. Self-report measures comprised visual analog scales (VAS) for rapid time-course ratings of general drug effects and four closed-eye visual categories (patterns, things, beings, scenes), plus the Altered States of Consciousness (ASC) Visionary Changes subscales and the Hood Mysticism questionnaire, administered 7.5 hours after dosing. For rapidly changing measures the analysis used each participant's maximum post-dose change (Emax). The tilt illusion task probed putative changes in early visual cortical excitation by measuring shifts in perceived orientation of a central sine-wave grating when embedded in oriented surrounds. Points of subjective verticality (PSV) were estimated using interleaved staircases for multiple surround orientations and the tilt-illusion magnitude was computed as the difference between minimum and maximum PSV after baseline correction. The contour integration task assessed perceptual organisation using closed chains of Gabor elements embedded in background noise; difficulty was manipulated by orientation jitter and thresholds (75% accuracy) were estimated from psychometric functions. The object recognition task assessed recognition of phase-degraded images presented with true, false or no semantic cues; images were progressively deblurred until the participant reported recognition. Primary outcomes were recognition accuracy and the degradation level at identification, analysed separately by cue type. Statistical analyses used mixed-effects models in R with drug condition as a fixed effect and participant as a random effect. Sequence and sequence-by-condition terms were initially included to control for order effects. Repeated measures were reduced to Emax for analysis where appropriate. After omnibus F-tests, pairwise comparisons used Tukey post hoc tests. The authors report handling of a few missing or unusable data points arising from participant misunderstanding or technical errors in task runs.

MDA was generally well tolerated and produced robust subjective and physiological effects consistent with hallucinogen action. On VAS measures of general drug effects, MDA significantly increased maximum ratings for any drug effects (t = 18.624, p < 0.001), good drug effects (t = 9.769, p < 0.001), high (t = 13.35, p < 0.001) and, to a lesser extent, bad drug effects (t = 2.366, p = 0.0272). Approximately half the participants reported notable closed-eye visions during the MDA session; CEVs were largely absent under placebo. MDA significantly increased Emax for closed-eye patterns (t = 4.437, p < 0.001), closed-eye objects (t = 3.883, p < 0.001), closed-eye beings (t = 2.54, p = 0.019) and closed-eye scenes (t = 2.79, p = 0.011). Scores on the ASC main scales and the Hood Mysticism questionnaire were also significantly elevated after MDA, including all Visionary Changes subscales. On the tilt illusion task, a planned within-subjects analysis did not find a significant effect of dosing condition on illusion magnitude (F1,9 = 2.515, p = 0.147). However, the authors detected a significant sequence effect (F1,9 = 6.658, p = 0.030) and therefore analysed session-one data alone; in that between-subjects comparison dosing condition predicted a larger tilt illusion under MDA (F1,9 = 7.495, p = 0.023). The tilt-illusion magnitude did not correlate with peak CEVs. Contour integration thresholds were not predicted by dosing condition alone (F1,10 = 0.409, p = 0.537). When peak CEVs were included as a predictor, there was a significant main effect of CEVs (F1,9 = 9.385, p = 0.014) and a significant interaction between CEVs and dosing condition (F1,9 = 17.972, p = 0.022), indicating that individuals with larger CEVs had poorer contour-detection performance and showed greater MDA-related impairment. In the object recognition task there were main effects of cue type (F2,52 = 6.035, p = 0.004) and dosing condition (F1,52 = 6.058, p = 0.017) on accuracy, with true cues improving accuracy by an estimated 7.2% and MDA reducing accuracy by an estimated 4.9% relative to placebo. Adding image degradation level revealed that participants on MDA were more impaired by increased degradation (interaction F1,46 = 6.693, p = 0.013). Collapsing cue types, peak CEVs and dosing condition both predicted accuracy, with significant main effects of CEVs (F1,10 = 20.337, p = 0.001) and condition (F1,9 = 8.736, p = 0.016) and a significant interaction (F1,9 = 11.499, p = 0.008). The authors report that each percent increase in CEVs was associated with approximately a 0.2% decrease in accuracy on MDA and 0.1% decrease on placebo, and that these relationships were specific to visual effects rather than to general drug effects. Missing or unusable data affected a few runs (one participant excluded from contour analysis, one tilt session unusable, and one missing object-recognition session).

Baggott and colleagues interpret their findings as confirming that MDA produces prototypical hallucinogen effects in humans, including increases in closed-eye visual phenomena and indices of mystical-type experience on standard questionnaires. The authors regard MDA as a useful experimental tool for studying hallucinogen-induced visions, while noting its complex pharmacology and limited prior human research. A key observation was that larger CEVs were associated with poorer performance on measures of perceptual organisation (contour integration and object recognition), and that those with more intense CEVs showed greater MDA-related impairment. The researchers argue this pattern is consistent with the hypothesis that reduced sensory fidelity or impaired perceptual organisation increases susceptibility to drug-induced visual phenomena. They note that the acute contour-integration effects under MDA resemble prior findings with the NMDA antagonist ketamine, supporting a cross-drug link between impaired early visual processing and hallucination-like experiences. Evidence for increased occipital cortical excitation, as indexed by the tilt illusion, was equivocal. A between-subjects comparison limited to first sessions suggested a dosing effect, but planned within-subject analyses did not replicate this and the tilt-illusion change was not related to CEV magnitude or self-reported geometric visuals; the authors therefore consider the tilt-illusion result tentative and possibly influenced by residual drug effects. With respect to cognitive, top-down mechanisms, the object recognition task showed that true cues improved recognition and that MDA reduced overall accuracy and increased sensitivity to image degradation. However, the study did not find a significant interaction between cue type and dosing condition, and the task did not cleanly separate criterion shifts from sensitivity changes. Consequently, the investigators could not confirm or refute a role for strengthened top-down influences in producing MDA-induced visions. The authors acknowledge several limitations that constrain interpretation: participants were experienced users, which may affect expectancy and generalisability; dosing sessions occurred on consecutive days, producing sequence and possible residual effects that complicated within-subject comparisons; and some tasks could not disentangle certain perceptual from decisional factors. They recommend future studies using signal-detection paradigms or noise masking to partition sources of altered efficiency and suggest repeating these paradigms with other serotonergic hallucinogens such as LSD or psilocybin to clarify links between neuropharmacology and phenomenology. Overall, the study supports an association between visual hallucination-like experiences and impaired perceptual processing, while providing only limited evidence for increased cortical excitation or altered top-down efficacy as primary mechanisms.