Additive Effects of 3,4-Methylenedioxymethamphetamine (MDMA) and Compassionate Imagery on Self-Compassion in Recreational Users of Ecstasy

Ecstasy (street MDMA) is widely used recreationally and has well-documented prosocial subjective effects, including heightened interpersonal understanding and compassion. Controlled laboratory work with MDMA has reported increases in emotional empathy and reductions in some aspects of cognitive empathy, and a number of neurobiological mechanisms have been proposed to explain these effects, notably MDMA-induced increases in central oxytocin. Behavioural strategies aimed at increasing self-directed affiliation, such as compassionate imagery (CI) drawn from compassion-focused therapy, can produce similar changes in self-attitudes, but individual differences (for example, fear of compassion or insecure attachment) can limit their effectiveness. Kamboj and colleagues note prior naturalistic findings suggesting that ecstasy and CI can both increase state self-compassion and that their effects may summate, but emphasise that earlier work did not chemically verify ecstasy composition and so could have been influenced by adulterants or by uncontrolled aspects of "set and setting". This study set out to replicate and extend previous naturalistic findings while chemically verifying participants' self-sourced ecstasy. The main hypotheses were that recreational MDMA use and CI would each increase state self-compassion and reduce self-criticism, and that combined administration (MDMA + CI) would produce additive effects on these self-attitudes. A secondary prediction was that MDMA would enhance emotional arousal in response to compassionate facial expressions, assessed using an emotional empathy task with both basic (anger, happiness) and complex (criticism, compassion) facial stimuli. The investigators emphasised that confirming MDMA content of participants' tablets would allow clearer attribution of observed effects to MDMA itself.

Recreational ecstasy users were recruited by word-of-mouth and screened by telephone; those with serious ongoing medical or psychiatric illness, pregnant or breastfeeding women were excluded. Twenty-five participants began and completed the protocol; chemical analysis retained data from 20 participants (12 men, 8 women) whose samples contained only MDMA (n = 18) or MDMA plus ≤ 33% glucose (n = 2). Participants were predominantly polydrug users and reported prior regular MDMA experience (median 4 years; median frequency 1/month). Sixteen participants were smokers and most (n = 17) did not practice meditation. Participants received £30 for participation. A naturalistic, within-subjects crossover design was used. Each participant completed two home-based testing sessions 6–14 days apart: one session in which they consumed their usual self-sourced ecstasy dose ~1 hour before a standardised guided compassionate imagery (CI) exercise (MDMA + CI session), and one session in which no drug was taken prior to CI (CI-only session). Session order was approximately balanced (9 of the 20 completed MDMA + CI first). Assessments occurred at three time points: T1 (baseline), T2 (~1 h after T1; on the MDMA + CI session this corresponded to near-peak drug effect) and T3 (~20 min after T2, i.e. after CI). Between T1 and T2 participants read or listened to music; CI was delivered after T2 via standardised audio instructions including rhythmic breathing and imagery of an ideal compassionate being. Primary psychometric outcomes were state self-compassion and self-criticism measured by the scenario-based Self-Compassion and Criticism Scale (SCCS). Secondary measures included the Types of Positive Affect Scale (TPAS), Beck Depression Inventory-II (BDI-II), the Revised Adult Attachment Scale (AAS), and an emotional empathy task (EAT-SAM) that recorded subjective arousal and valence in response to images of basic emotions (anger, happiness drawn from NimStim) and custom-generated complex interpersonal expressions (criticism, compassion). Physiological assessment comprised ambulatory ECG-derived heart rate. Urine drug screens and breathalyser tests documented recent substance use. Chemical verification of ecstasy samples (~10 mg retained prior to consumption) was performed using 1H NMR spectroscopy and, when needed, additional 13C and 2D NMR and ESI-MS to identify constituents. Statistical analyses used repeated-measures ANOVAs (Time × Session) for subjective and SCCS outcomes and three-way ANOVAs (Time × Session × Emotion) for EAT-SAM arousal, with separate analyses for basic and complex emotion pairs. ANCOVAs tested moderation by attachment subscales (AAS closeness and dependence). Data were checked for normality and outliers; parametric tests were applied. Effect sizes reported included partial eta-squared for ANOVAs and Cohen's d (adjusted for paired observations) for planned contrasts. Key analyses were re-run excluding an outlier who consumed 0.25 g MDMA and excluding the three participants who insufflated their dose; results were reported as robust to these exclusions.

Twenty participants contributed chemically verified MDMA-only data (mean age 28.45 ± 6.16 years; mean BDI-II 9.50 ± 6.53). Urine drug screens showed some positive results across sessions (e.g. THC, cocaine, occasional MDMA), reflecting recent drug use in this polydrug sample; breathalyser readings were 0.00% on both sessions. Visual analogue mood and symptom scales showed multiple Time × Session interactions consistent with expected MDMA effects (increases in energetic, jaw clenching, muscle tension, sensitivity to colour, 'high' and alertness, and a decrease in hunger) between T1 and T2/T3 on the MDMA + CI session. Heart rate increased significantly in the MDMA + CI session relative to the CI-only session, by 14.15 ± 18.93 beats/min at T2 and 14.63 ± 19.39 beats/min at T3 (Session × Time interaction F(1.4, 23.5) = 9.034, p = 0.003, ηp2 = 0.347). On the SCCS self-criticism subscale there were main effects of Time (F(2,38) = 26.745, p < 0.001, ηp2 = 0.585) and Session (F(1,19) = 6.881, p = 0.017, ηp2 = 0.266) but the Time × Session interaction did not reach significance (F(2,38) = 2.584, p = 0.089). By contrast, the SCCS self-compassion subscale showed main effects of Time (F(1.5,28.1) = 10.285, p = 0.001, ηp2 = 0.351) and Session (F(1,19) = 6.592, p = 0.019, ηp2 = 0.258) that were subsumed by a significant Time × Session interaction (F(2,38) = 5.794, p = 0.006, ηp2 = 0.351). Follow-up contrasts indicated: on the CI-only session there was no change from T1 to T2 (extended baseline) but a modest increase in self-compassion from T2 to T3 after CI (p = 0.023, d = 0.355). On the MDMA + CI session self-compassion increased significantly from T1 to T2 (post-MDMA; p = 0.008, d = 0.396) and showed a further small increase from T2 to T3 (post-CI; p = 0.015, d = 0.191). Scores at T1 did not differ between sessions, but at T3 state self-compassion was higher on the MDMA + CI session than on the CI-only session (p = 0.003, d = 0.301). No Session × Time interactions emerged for the TPAS positive-affect subscales (active, relaxed, content/warm), suggesting specificity of effects to self-compassion rather than broad increases in positive affect. Analyses testing attachment moderation found no significant Time × Session × AAS-closeness effect; using AAS-dependence as a covariate produced a trend-level Time × Session × Attachment interaction (F(2,36) = 3.039, p = 0.06, ηp2 = 0.144), with descriptive evidence that higher dependence scores were associated with larger self-compassion increases. On the EAT-SAM, basic emotions showed only main effects of Emotion (arousal: anger > happiness; valence: happiness > anger) with no Session or Time effects. For the complex emotion pair there was a Time × Session × Emotion interaction on arousal (F(1,19) = 5.155, p = 0.035, ηp2 = 0.213). Subsequent tests revealed a strong Time × Session interaction for critical faces (F(1,19) = 16.429, p = 0.001, ηp2 = 0.464): at T2 arousal to critical faces was higher on the MDMA + CI session relative to the CI-only extended baseline (p < 0.001, d = 0.768), and on the CI-only session arousal to critical faces increased from T2 to T3 after CI (p = 0.019, d = 0.391). There was no evidence of an additive increase in arousal to critical faces between T2 and T3 on the MDMA + CI session. No significant correlations were found between EAT-SAM ratings to critical faces and concurrent self-compassion or self-criticism scores.

Kamboj and colleagues interpret these results as an extension of earlier naturalistic observations: both recreational MDMA use and a brief guided compassionate imagery exercise produced small-to-medium increases in state self-compassion when administered separately, and the combination yielded a further, statistically significant but modest, increase. The authors highlight that the current study's chemical verification of participants' ecstasy strengthens the attribution of these effects to MDMA itself, relative to prior work where tablet composition was unknown. They note that changes in self-compassion occurred without parallel Session × Time effects on three distinct types of positive affect, suggesting the effects were not merely a generalised mood increase or demand artefact. In terms of interpersonal processing, the study found that emotional arousal in response to critical faces increased following MDMA (at T2) and following CI (T2 to T3 on the CI-only session), whereas no MDMA- or CI-related enhancement of arousal to compassionate faces was observed. This pattern departs from some previous experimental reports that emphasised MDMA-related increases in empathy to positive stimuli; the investigators suggest methodological differences (use of facial rather than situational stimuli, and novel critical-face stimuli) may explain the divergence. Mechanistically, the authors propose the oxytocinergic system as a plausible shared substrate for the observed increases in self-compassion, given oxytocin's role in affiliation and its hyperstimulation by MDMA, but acknowledge autonomic pathways differ for MDMA (sympathetic activation) versus compassion practices (parasympathetic engagement), so a single autonomic explanation is unlikely. The authors are cautious about therapeutic implications. They stress that the findings are preliminary, derived from a naturalistic, non-placebo-controlled design, and that generalisability to MDMA-naive individuals or clinical populations is unknown. Key limitations discussed include the naturalistic setting and attendant expectancy effects, inability to blind participants, variability in self-chosen MDMA doses (one outlier at 0.25 g), presence of recent other drug use detectable in urine screens, and absence of additional control conditions (for example an MDMA-only session or an active control behavioural task). Kamboj and colleagues therefore recommend that future work employ randomised, double-blind, controlled laboratory designs to test whether MDMA combined with compassion-oriented psychosocial strategies can produce lasting therapeutic gains, and to clarify mechanisms and boundary conditions for potential clinical application.