'Ecstasy' (MDMA) on Processing of Facial Expressions

Drug and Alcohol Dependence 76 (2004) 297–304

The acute and sub-acute effects of ‘ecstasy’ (MDMA) on processing of facial expressions: preliminary ﬁndings

Rosa Hoshi∗, Jatinder Bisla, H. Valerie Curran

Clinical Psychopharmacology Unit, Sub-Department of Clinical Health Psychology, University College London, Gower Street, London WC1E 6BT, UK Received 6 April 2004; received in revised form 25 June 2004; accepted 30 June 2004

Abstract

Rationale: There is evidence that serotonergic processes may modulate the processing of fearful facial expressions. It is therefore possible that the recreational drug ‘ecstasy’ (MDMA), which has marked serotonergic effects, may affect people’s ability to recognise human facial expressions portraying fear. Objective: The present study therefore aimed to determine whether ecstasy users differed from controls in fear recognition at two time points: shortly after taking the drug and a few days later. Methods: Sixteen ecstasy users and 21 controls were compared on a facial expression recognition task involving the 6 basic emotions (happiness, surprise, sadness, anger, fear and disgust) and on self-ratings of mood on the night of drug use (day 0) and 4 days later (day 4). Results: In recognising fearful facial expressions, ecstasy users were more accurate than controls on day 0 but less accurate than them on day 4 when compared with their overall ability to recognise other basic emotions. Accuracy of fear recognition on day 4 was negatively correlated with both years of ecstasy use and number of ecstasy tablets taken on a typical session. On self-rated aggression scales, ecstasy users scored lower than controls on day 0 and higher on day 4. Conclusions: These results support the notion that 5-HT plays a role in modulating the recognition of fearful facial expressions. Increased accuracy of fear recognition may relate to 5-HT release following ecstasy use on day 0, and decreased accuracy may reﬂect subsequent depletion of 5-HT mid-week. © 2004 Elsevier Ireland Ltd. All rights reserved.

Keywords: MDMA; Ecstasy; Serotonin; Fear recognition; Facial expressions

1. Introduction through these systems, MDMA’s main action is on the serotonergic (5-HT) system. After binding to the 5-HT trans- ±3,4-Methylenedioxymethamphetamine (MDMA, ‘ec- porter, the MDMA molecule is taken into the presynaptic stasy’) is used recreationally for its acute subjective effects cell where it stimulates the release of stored 5-HT and pre- that include euphoria, insightfulness, energy and feelings of vents its reuptake (Green et al., 1995). MDMA administra- closeness to others (Vollenweider et al., 1998; Harris et al., tion causes short-term attenuation of tryptophan hydroxylase, 2002). These subjective effects, described by many users as thus inhibiting the synthesis of new 5-HT. This may have im- feelings of empathy, appear unique to MDMA and have led plications for recreational users, as the possible depletion of to the suggestion that MDMA should be classiﬁed as a new brain 5-HT in the days following MDMA use may affect a class of psychoactive compounds known as ‘entactogens’ variety of functions modulated by 5-HT. (Nichols, 1986). The recognition of facial expressions is a “crucially im- MDMA causes the release of dopamine (DA) and nora- portant” element of social interaction (Harmer et al., 2003a). drenaline (NE), but whereas other stimulant drugs act mainly The six basic emotions—happiness, surprise, sadness, anger, fear and disgust—are consistently recognised across cultures ∗ Corresponding author. Tel.: +44 207 679 5938; fax: +44 207 916 1989. (Ekman, 1992) and there is some evidence that each may E-mail address: [email protected] (R. Hoshi). be modulated by different neural substrates (LeDoux, 2000).

Manipulation of different neurotransmitters seems to affect and Lader, 1986), impulsivity self rating scale (ISRS; Bond processing of facial expressions differentially (e.g. Blair and and Lader, 1974) and a scale of subjective effects which cov- Curran, 1999). Harmer and colleagues have assessed the ef- ered common effects of recreational drugs including those fect of various 5-HT manipulations on recognition of facial specific to MDMA. A modified Beck depression inventory expressions. They found that on the one hand, increased (BDI; Beck, 1978) was used to assess how participants had release of 5-HT through either intravenous citalopram or been feeling over the last 3 days (Curran and Travill, 1997). tryptophan administration leads to improved recognition of fearful facial expressions (Harmer et al., 2003a; Attenburrow 2.4. Recognition of facial expressions (days 0 and 4) et al., 2003); on the other hand, reducing 5-HT by tryptophan depletion impaired recognition of fear (Harmer et al., 2003b). The participants performed the facial expression recogni- Recognition of facial expressions has not, to our knowl- tion task on both days of the study. The task used the stimuli edge, ever been assessed in MDMA users. Thus, in light of from the facial expressions of emotions: stimuli and tests the findings described above, the current study was designed (FEEST; Young et al., 2002). There were 30 stimuli featur- to address the hypothesis that enhanced 5-HT following acute ing a male face showing the 6 basic emotions—happiness, MDMA ingestion would selectively improve recognition of surprise, sadness, anger, fear and disgust—from the Ekman fearful facial expressions and that several days later depletion and Friesen (1976) series. These basic emotions are used to of 5-HT would lead to a selective impairment of fear recog- create stimuli that are morphed from one emotion to another nition. In addition, we expected to replicate previous studies in five stages (10, 30, 50, 70, 90%). The expressions morphed that demonstrated increased self-rated aggression and depres- are anger to happiness, happiness to surprise, surprise to fear, sion 4 days after ecstasy use (Verheyden et al., 2002; Curran fear to sadness, sadness to disgust and disgust back to anger. et al., 2004). All stimuli involved the face JJ (Ekman and Friesen, 1976). The task was performed on lap top computers and con- sisted of six blocks. In each block, the 30 morphed stimuli 2. Materials and methods were presented in a pseudo-random order. Constraints were in place to ensure that no more than two stimuli with the same 2.1. Design and participants majority emotion appeared together. Each stimulus appeared An independent group, repeated measures design was used on the screen for 500 ms. The task was specially programmed to compare ecstasy users with controls on two test sessions: with a response facility whereby the six emotions were ar- the night of drug use (day 0) and 4 days later (day 4). Both ranged in an equilateral hexagon in the centre of the right the ecstasy users and the control participants were recruited hand side of the screen next to the faces which were pre- in clubs and parties using the snowball sampling technique sented on the left hand side of the screen. This arrangement (Solowij et al., 1992). In this way all participants were re- was such that each emotion was equidistant from the central cruited from the same social settings and increased the like- cursor base. Participants were asked to choose the emotion lihood of matching groups on the use of other drugs. The that corresponded with the facial expression by clicking on it study was approved by the institutional ethics committee and with the mouse, as quickly and accurately as they could. The all participants gave written informed consent both on day 0 position of the emotions around the hexagon changed on each and on day 4. occasion to avoid any response bias, and the mouse cursor re- turned automatically to the centre of the hexagon before each 2.2. Procedure stimulus was presented. This response facility was designed to both lessen the working memory load required to remem- On the evening of day 0 participants were taken individ- ber six different response keys and ensure that the response ually to a quiet room where they were first given an infor- to each emotion required equal motor movement. Both re- mation sheet. If they were willing to participate they were sponse and reaction time were automatically recorded, and asked for written consent. They then completed the assess- recognition accuracy and reaction times used in the analysis ments detailed below. Arrangements were made for the next were calculated from the correct responses to the stimuli that test session 4 days later, which was generally carried out at had a dominant percentage (90 or 70%) of each emotion. the participant’s home. Informed consent was obtained once Each participant’s pulse rate was taken at both test ses- again on day 4, then the day 0 assessments were repeated sions. along with trait measures of mood and a drug use history interview. 3. Additional assessments on day 4 2.3. Mood assessments (days 0 and 4) 3.1. Trait measures and pre-morbid IQ Current mood state was assessed on both days with the following visual analogue scales: mood rating scale (MRS; The hospital anxiety and depression scale (HADS; Bond and Lader, 1974), aggression rating scale (ARS; Bond Zigmond and Snaith, 1983) was used to index trait anxiety R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304 299 and depression, the aggression questionnaire (AQ; Buss and (6 males, 15 females). There was a significant difference be- Perry, 1992) was used to index trait aggression, and the Bar- tween the number of males and females in each group (χ2 = ratt impulsivity scale (BIS; Barratt and Patton, 1983)was 4.26, P = 0.039). Due to this imbalance, gender was used as a used as a measure of trait impulsivity. The ‘Spot the Word’ covariate throughout the analyses. There were no group dif- test (Baddeley et al., 1992) was used to index pre-morbid ferences in age, pre-morbid IQ (Spot the Word), trait aggres- IQ as this measure correlates highly with the national adult sion (AQ), impulsivity (BIS), anxiety or depression (HADS) reading test (NART; Nelson and O’Connoll, 1978). (Table 1). All but two participants in the control group were either undergraduate students or graduates in full time em- 3.2. Drug use history ployment. In the ecstasy group, participants reported using ecstasy Amount, frequency and duration of use of all psychotropic for an average of 3.31 (2.18) years. They took ecstasy on drugs were recorded as well as details of drug use on day 0, an average of 2.63 (1.41) days in a typical month, taking 4 and the period between the 2 days. Demographic informa- 2.25 (1.05) tablets in a typical session. Seven of the control tion (gender, age, educational background and employment group had previously tried ecstasy, but none were regular status) was also noted on day 4. users. They did, however, regularly use other recreational drugs such as cannabis. There were no group differences in 3.3. Statistical analysis measures of either alcohol or cannabis use (Table 1). How- ever, two measures of cocaine use differed significantly be- A2× 2 repeated measures ANOVA with group (ecstasy tween the groups: ecstasy users had greater years of use (U = versus controls) as a between subject factor and day (day 101.50, P = 0.024, two-tailed) and frequency of use [F(1,35) 0 versus 4) as a within-subject factor was used on VAS and = 4.84, P = 0.035). Reported regular (>1 day per month) use BDI scores assessed on both days. The 13 variables that make of other drugs included amphetamines (one ecstasy user) and up the ARS showed high internal consistency on both day 0 ketamine (two ecstasy users). (Cronbach’s α = 0.88) and day 4 (α = 0.90); the ISRS also On day 0, the ecstasy users took an average of 3.06 (1.12) showed high internal consistency on both days (day 0, α = ecstasy tablets (Table 2). All the participants had drunk al- 0.80; day 4, α = 0.89). Thus, a mean aggression and a mean cohol; two ecstasy users and three controls had smoked impulsivity score for each participant was used in subsequent cannabis. There were no significant differences in quanti- analysis. As the subjective effects scale comprises 26 scales, ties of alcohol and cannabis consumed on day 0 (Table 2). an alpha level of 0.01 was adopted in order to reduce the Only 1 control had used cocaine, compared to 10 ecstasy possibility of type I errors. One-way ANOVAs were used users, and there was a significant difference in the number of to compare groups in use of alcohol, cannabis and cocaine, ‘lines’ of cocaine used (U = 75.50, P < 0.01, two-tailed). apart from the comparison of years of cannabis use, years of Between the 2 test sessions 5 ecstasy users had smoked cocaine use and amount of cocaine used on day 0 where a cannabis and 12 ecstasy users and 10 controls had drunk al- Mann–Whitney U-test was used due to non-parametric data. cohol. In addition, two ecstasy users and two controls had A2× 2 × 6 repeated measures ANOVA was used to analyse drunk between 2 and 4 units alcohol prior to testing on the facial expression recognition task, with group (ecstasy day 4. versus controls) as a between subject factor, and emotion (happiness, surprise, sadness, anger, fear and disgust) and day Table 1 (day 0 versus 4) as within-subject factors. As the assumptions Group means (S.D.) for age, Spot the Word test, aggression questionnaire of sphericity were violated, a Greenhouse–Geisser correction (AQ) score, Barratt impulsivity scale (BIS) score, HADS-D (depression), was used. The data from two participants in the ecstasy group HADS-A (anxiety) and measures of alcohol and cannabis use was not analysed due to computer failure on day 0. As a signif- Ecstasy users Controls icant gender differences was found between the groups (see Age 21.94 (2.98) 20.95 (2.11) Section 4.1), gender was used as a covariate throughout the Spot the Word 46.38 (4.00) 48.05 (4.03) analyses. The relationship between recognition of fearful fa- AQ 68.50 (12.05) 67.62 (14.61) cial expressions and ecstasy use was explored using Pearson BIS 55.06 (12.35) 53.90 (15.72) HADS-D 3.06 (1.88) 2.57 (2.20) product–moment correlations. All data were analysed using HADS-A 7.06 (1.88) 6.95 (3.60) SPSS for Windows versions 11.0. Alcohol frequency (days per month) 14.75 (7.36) 11.86 (6.79) Alcohol typical dose (units per session) 7.81 (2.37) 9.10 (3.71) Alcohol length of use (years 6.75 (3.04) 5.57 (1.60) 4. Results Cannabis frequency (days per month) 4.56 (5.09) 4.29 (8.41) Cannabis typical dose (joints) 1.03 (1.16) 0.52 (1.36) Cannabis length of use (years) 3.19 (4.25) 0.88 (1.84) 4.1. Demographics Cocaine frequency (days per month) 2.06 (1.88)∗ 0.86 (1.46) Cocaine typical dose (lines) 4.31 (4.45) 1.90 (3.95) There were 37 participants aged between 20 and 32 years Cocaine length of use (years) 2.00 (2.07)∗ 0.62 (1.07) old, 16 ecstasy users (10 males, 6 females) and 21 controls ∗ Significant at P < 0.05. 300 R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304

Table 2 Means (S.D.) of ecstasy, alcohol and cannabis use before testing, after testing and total on day 0 Ecstasy users Controls

Before After Total Before After Total Ecstasy (tablets) 2.72 (1.08) 0.34 (0.60) 3.06 (1.12) 0.00 0.00 0.00 Alcohol (units) 10.88 (5.97) 1.69 (2.30) 12.56 (6.12) 10.71 (3.87) 2.00 (2.30) 12.71 (4.37) Cannabis (joints) 0.00 0.19 (0.54) 0.19 (0.54) 0.45 (1.47) 0.14 (0.36) 0.60 (1.53) Cocaine (lines) 0.81 (1.05) 0.56 (0.95)∗∗ 1.38 (1.30)∗∗ 0.19 (0.87) 0.00 0.79 (0.87) ∗∗ Signiﬁcant at P < 0.01.

4.2. Recognition of facial expressions

A significant emotion × day × group interaction was found for accuracy [F(5,160) = 3.260, P = 0.019], as well as significant main effect of emotion [F(5,160) = 14.875, P < 0.0001] and day [F(1,32) = 13.193, P = 0.001] (Table 3). Re- peated measure ANOVAs with day as a within subject variable, group as the between subjects variable and gender as a covariate were carried out for each emotion separately. Trends towards significant day × group interactions were found for fear [F(1,32) = 3.185, P = 0.084] and disgust [F(1,32) = 3.517, P = 0.07]. Disgust also showed a significant day × gender interaction [F(1,32) = 4.696, P = 0.038] and a main effect of day [F(1,32) = 5.474, P = 0.026]. No emotion × gender interactions were found. No simple effects were found when Fig. 1. Accuracy of recognition of fearful facial expressions compared to all comparing groups on each individual emotion on each day other emotions by ecstasy users and controls on day 0 and 4. separately. As the hypothesised difference was expected on 4.3. Self-rated scales recognition of fear, and trends toward significant interactions were found for fear and disgust, recognition data for these 4.3.1. Aggression rating scale (ARS) two emotions were individually compared to the mean ac- A significant group × day interaction was found on state curacy of all the other five emotions. Comparing fear to all × × aggression ratings [F(1,34) = 14.164, P = 0.001]. Ecstasy other emotions yielded a significant emotion day group users’ ratings were lower than controls on day 0 but higher interaction [F(1,32) = 4.293, P = 0.046], a significant main than controls on day 4 (Fig. 2). There were no significant effect of emotion [F(1,32) = 13.938, P = 0.001] and of day gender effects. [F(1,32) = 9.898, P = 0.004]. The ecstasy users were more accurate than controls at recognising fear on day 0 and less 4.3.2. Mood rating scale (MRS) accurate than controls at recognising fear on day 4 (Fig. 1), The MRS yields three factors: sedation, discontented- although no significant posthoc differences were found. The ness and anxiety. A group × day interaction was found for groups did not differ in the mean recognition of the other five emotions. No significant interactions were found when comparing disgust to all other emotions. No significant interactions or main effects were found for the analysis of reaction time.

Table 3 Percentage (S.D.) correct for each emotion of days 0 and 4 Day 0 Day 4

Ecstasy users Controls Ecstasy users Controls Happiness 85.7 (21.35) 89.3 (13.86) 96.1 (5.53) 94.4 (8.78) Surprise 72.0 (20.57) 76.2 (16.83) 81.5 (6.88) 74.6 (13.62) Sadness 61.0 (25.77) 63.7 (23.28) 78.0 (11.14) 72.6 (17.56) Anger 41.7 (23.80) 33.9 (25.56) 70.5 (15.63) 56.3 (28.19) Fear 50.6 (23.68) 43.8 (22.23) 54.5 (20.64) 63.9 (20.34) Fig. 2. Mean aggression scores on day 0 and 4 for the ecstasy and control Disgust 64.6 (23.67) 61.3 (29.00) 63.4 (19.14) 72.4 (20.69) groups. R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304 301

Table 4 Group means (S.D.) on days 0 and 4 for the aggression rating scale (ARS), mood rating scale factors, Beck depression inventory (BDI), impulsivity self rating scale (ISRS), pulse rate and subjective effects scales Ecstasy users Controls

Day 0 Day 4 Day 0 Day 4 MRS-sedation 37.92 (18.13) 45.28 (15.35) 50.38 (14.31) 34.89 (15.14) MRS-discontentedness 19.26 (12.42) 36.20 (14.59) 24.82 (13.58) 30.79 (18.25) MRS-anxiety 35.50 (23.28) 30.09 (11.52) 40.60 (21.00) 27.23 (19.40) BDI 4.88 (3.58) 5.94 (3.17) 5.52 (5.93) 4.81 (3.89) ISRS 51.90 (16.72) 52.63 (16.30) 58.93 (14.99) 43.69 (19.16) Pulse rate 99.14 (14.80) 72.86 (8.07) 79.81 (8.02) 71.24 (6.56) Teeth grinding 43.75 (28.25) 8.09 (14.77) 14.31 (23.07) 11.24 (19.91) Jaw clenching 46.81 (29.48) 8.63 (17.16) 11.62 (18.49) 8.40 (13.65) Dry mouth 57.03 (29.07) 21.69 (26.12) 29.98 (25.91) 18.31 (23.98) Energy 59.28 (31.31) 35.25 (15.70) 42.64 (28.32) 44.45 (22.70) Euphoria 67.31 (16.53) 19.34 (15.76) 44.67 (27.89) 15.19 (15.94) Open to others 72.31 (14.54) 37.56 (18.65) 59.57 (27.65) 43.76 (19.19)

alert/sedated [F(1,34) = 12.516, P = 0.001] showing that ec- 4.4. Pulse rate stasy users were less sedated than controls on day 0 and more sedated than controls on day 4 (Table 4). There was no main Two participants in the ecstasy group did not have their effect of group or day. A significant day × group interac- pulse taken on day 0, thus their pulse data on day 4 was also tion was found for contented/discontented [F(1,34) = 4.866, excluded. A significant group × day interaction [F(1,32) = P = 0.034, as well as a day × gender interaction [F(1,34) 22.072, P < 0.0001] as well as a significant main effect of = 5.377, P = 0.027]. In both groups males were less con- both day [F(1,32) = 65.876, P < 0.0001] and group [F(1,32) tented that females on day 0 and more contented on day 4. = 11.622, P = 0.002] reflected higher pulse rates in ecstasy The calm/anxious scale showed no significant interactions or users compared to controls only on day 0 (Table 4). There main effects (Table 4). were no significant gender interactions.

4.3.3. Beck Depression Inventory (BDI) 4.5. Correlations There were no significant interactions or main effects on BDI scores (Table 4). Accuracy of fear recognition and self-rated aggression were correlated with self-reported ecstasy use. Within the 4.3.4. Impulsivity self rating scale ecstasy group, accuracy of fear recognition on day 4 was neg- A significant group × day interaction was found on state atively correlated with both years of ecstasy use (r = −0.54, P impulsivity self-ratings [F(1,34) = 5.671, P = 0.023]. Although = 0.048) and number ecstasy tablets taken on a typical session the ecstasy users’ ratings were similar on both test days, the (r = −0.58, P = 0.031). As the groups differed significantly controls scored higher than the ecstasy group on day 0 and on cocaine use on the night, this measure was also correlated relatively lower on day 4 (Table 4). There were no significant with fear recognition on day 0 (r = −0.306, P = 0.287) and gender effects. day4(r = −0.015, P = 0.959), but no significant correlations emerged. Significant correlations were found between 4.3.5. Subjective effects scale (SES) years of cocaine use and ecstasy use (r = 0.83, P < 0.001) At P < 0.01 level, four of the scales showed a significant and the number of days that cocaine and ecstasy were used group × day interaction: energy [F(1,34) = 8.852, P = 0.005], in a typical month (r = 0.67, P = 0.005). euphoria [F(1,34) = 9.209, P = 0.005], teeth grinding [F(1,34) = 11.829, P = 0.002] and jaw clenching [F(1,34) = 15.470, P < 0.0001], with ecstasy users showing higher ratings than 5. Discussion controls on day 0, but similar levels on day 4. In addition, two scales showed a significant day × gender interaction: As predicted by the original hypothesis, there was some anxiety [F(1,34) = 9.642, P = 0.004] and agitated [F(1,34) = evidence that participants in the ecstasy-using group recog- 12.866, P = 0.001]. In both cases male participants scored nised a greater number of fearful facial expressions than con- higher than female participants on day 0, and lower on day 4. trols following acute MDMA use, but 4 days later, the reverse There were two scales that showed trends towards significant was found with ecstasy users recognising fewer fearful facial day × group interactions: dry mouth [F(1,34) = 5.064, P = expressions than controls. 0.031] and openness to others [F(1,34) = 5.049, P = 0.031]. The effect on fear recognition found in this study was In these two cases, ecstasy users scored higher on the scales subtle, emerging as a group and day interaction with fear than controls on day 0 and were similar on day 4 (Table 4). versus all other emotions. The subtlety of this effect may 302 R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304 derive from the relatively small number of participants and/or deficits in fear recognition independent of other expression the gender imbalance across groups. However, the results are such as anger (Sprengelmeyer et al., 1999) and masked pre- in accord with the findings of Harmer and colleagues that sentations of fearful facial expressions illicit increased blood increasing 5-HT release enhances recognition of fearful facial oxygen level-dependent fMRI signal in the amygdala, even expressions (Attenburrow et al., 2003; Harmer et al., 2003a) when fear processing is implicit and participants report only and decreasing 5-HT release impairs their ability to recognise seeing neutral faces (Whalen et al., 1998). fearful facial expressions (Harmer et al., 2003b). Two indices of ecstasy use—the number of years ec- These findings could be attributed to the serotonergic stasy has been used and the amount of tablets taken on changes caused by MDMA use, whereby on day 0 the efflux day 0—correlated negatively with the participants’ ability of stored 5-HT enhanced the recognition of fearful facial ex- to recognise fearful faces on day 4. Thus, the greater the pressions. On the other hand, the ecstasy users’ impairment in number of ecstasy tablets taken on day 0 the worse the par- recognition accuracy on day 4 accords with the notion of de- ticipants’ fear recognition accuracy on day 4. Similarly, the pleted 5-HT following MDMA use and suggests a reciprocal longer they had been using ecstasy, the worse their day 4 fear role for 5-HT in the processing of fear relevant cues. recognition accuracy score. These two correlations are clearly There were no significant group differences in reaction interrelated. As with most centrally acting drugs, tolerance times to correctly recognise fearful facial expressions. Thus, to ecstasy’s effects builds up over repeated use so that users changes in accuracy in recognising fearful expressions was generally increase the dose taken over time to attain the same not a by-product of any speed-accuracy trade off that might effect (Verheyden et al., 2003). Thus, longer use of a drug is be influenced by the stimulant effects of MDMA on the night associated with a higher dose per drug using session. Higher of use. Practice effects could also make the results difficult to dosage on day 0 would, in theory, lead to greater depletion interpret. Controls showed an improvement of 20.1% in fear of 5-HT mid-week, and in accordance with the hypothesis recognition from day 0 to day 4, which may largely reflect that depleted 5-HT leads to impairments in fear recognition, practice effect on the task. However, the ecstasy users’ fear higher dose on day 0 would translate to greater impairment recognition increased by only 3.9%. That the change in ec- on day 4. stasy users appears subtle probably reflects a more substantial Scores on self-rated aggression (ARS) replicated the re- relative decline if practice effects are taken into account. Prac- sults of two previous studies (Verheyden et al., 2002; Curran tice effects also emerged on happiness, sadness and anger. For et al., 2004) in that ecstasy users scored lower than controls all these emotions the increase in accuracy of recognition by on day 0 and higher on day 4. As no significant difference was ecstasy users from day 0 to day 4 was markedly greater than found between the groups on a state measure of aggression that for fear. (AQ), this suggests that mid-week increase in aggression is a Fear is thought to be the most difficult expression to recog- transient phenomena related to depletion of 5-HT following nise (Ekman and Friesen, 1976), and it has been argued acute MDMA use. This idea is supported by literature that that seemingly selective impairments in fear recognition in suggests that low levels of 5-HT are associated with increased patients with lesions is more a reflection of task difficulty aggression (Cleare and Bond, 1997) and that increases in 5- (Rapcsak et al., 2000). HT, through administration of SSRIs, leads to increases in It could be argued that our findings arise from impaired positive affiliate behaviour (Tse and Bond, 2002). learning following ecstasy use. However, there is as yet no ev- Unlike the findings of several previous studies (Curran and idence to suggest that acute MDMA administration impairs Travill, 1997; Parrott and Lasky, 1998; Verheyden et al., 2002; learning. In fact, augmentation of serotonin through acute Curran et al., 2004), no significant differences were found in citalopram administration has been found to improve mem- depression (BDI). The gender imbalance of the groups may ory consolidation in healthy volunteers (Harmer et al., 2002) have affected the present results, as female users have been and, as MDMA administration also leads to increased lev- shown to be more vulnerable to low mood following MDMA els of 5-HT available, there seems no reason to assume that use and female participants were in a minority in the present MDMA should cause impaired learning. On the other hand, study’s ecstasy-using group (Verheyden et al., 2002). there is evidence that ecstasy users off-drug have memory As with virtually any naturalistic investigation of recre- impairments compared to controls. However, as the ecstasy ational drug users, there are methodological considerations users improved on four out of the six emotions it is unlikely that limit the conclusions of this study (see Curran (2000) that either memory impairments or impaired learning caused for review). Although the groups were well matched on pre- by MDMA administration account for the results observed. morbid IQ, age, trait aggression, depression, impulsivity and The present findings are consistent with the view that 5-HT anxiety, as well as use of cannabis and alcohol, the dose or plays a wider role in fear-related processes. Several studies purity of MDMA taken is unknown. However, ecstasy users have found that serotonergic input to the amygdala plays an in the present study reported many of the ‘trademark’ acute important role in conditioning and fear responses in humans effects, such as teeth grinding and jaw clenching, that have and animals (for review see Davis and Whalen (2001)). The been found in lab-based studies of acute administration of amygdala has also been implicated in the processing of fear- MDMA (Harris et al., 2002; Vollenweider et al., 1998) and ful facial expressions. Patients with amygdala damage show their pulse rate was significantly higher on day 0, another R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304 303 factor that indicates stimulant consumption (Mas et al., 1999). for the functioning and well-being of recreational MDMA Although hair and urine analysis could have been used to ob- users. jectively confirm drug use on the night of the study, it would have been impractical given the setting of the study. The use of self-reported drug use histories is also problematic due to Acknowledgments their unreliability (Cooper et al., 2000). Both lifetime indices of cocaine use (frequency and years RH was supported by a Medical Research Council (UK) of use) and amount of the drug used on day 0 differed sig- studentship. nificantly between the groups. There were, however, no correlations between cocaine use and either fear recognition or self-reported aggression. There was also a gender imbalance References between the groups in the present study with more males in the ecstasy group. This could be of particular importance as Attenburrow, M.J., Williams, C., Odontiadis, J., Reed, A., Powell, J., Cowen, P., 2003. Acute administration of nutritionally sourced tryp- two out of the three studies by Harmer and colleagues used tophan increases fear recognition. Psychopharmacology 169, 104– only female participants (Attenburrow et al., 2003; Harmer et 107. al., 2003a) and the other found decreased recognition of fear Baddeley, A.D., Emslie, H., Nimmo-Smith, I., 1992. The Speed and Ca- following tryptophan depletion in female but not male par- pacity of Language Processing (SCOLP) Test. Thames Valley Test ticipants (Harmer et al., 2003b). However, covarying gender Company, Bury St. Edmonds. Barratt, E.S., Patton, J.H., 1983. Impulsivity: cognitive, behavioural and in the present study did not affect the pattern of significant psychophysiological correlates. In: Zuckerman, M. (Ed.), Biological results. There is evidence that females are more accurate at Basis of Sensation Feeling, Impulsivity and Anxiety. Erlbaum, Hills- recognising facial expressions (Thayer and Johnsen, 2000), dale, NJ. and evidence from functional magnetic resonance imaging Beck, A.T., 1978. The Beck Depression Inventory (BDI). The Psycho- indicates the possibility of gender differences in the neu- logical Corporation. Harcourt Brace Jovanovitz Inc., USA. Blair, R.J.R., Curran, H.V., 1999. Selective impairment in the recogni- ral correlates of facial expression recognition (Lee et al., tion of anger induced by diazepam. Psychopharmacology 147, 335– 2002). In addition, previous literature suggests there may be 338. gender differences in the acute, residual and long-term ef- Bond, A.J., Lader, M.H., 1974. The use of analogue scales in rating fects of MDMA (Liechti et al., 2001; Reneman et al., 2001; subjective feelings. Brit. J. Med. Psychol. 47, 211–218. Verheyden et al., 2002). Therefore, it would be important to Bond, A.J., Lader, M.H., 1986. A method to elicit aggressive feelings and behaviour via provocation. Biol. Psychiatry 22, 69–79. carry out a future study with a larger sample of female ecstasy Borrill, J.A., Rosen, B.K., Summerfield, A.B., 1987. The influence of users to explore these issues. alcohol on judgment of facial expression of emotion. Brit. J. Med. Interestingly, both groups showed the greatest improve- Psychol. 60, 71–77. ment in recognition for anger, with both groups recognising Buss, A.H., Perry, M., 1992. The aggression questionnaire. J. Pers. Soc. over 20% more on day 4. The results relating to anger could Psychol. 63, 452–459. Cleare, A.J., Bond, A.J., 1997. Does central serotonergic function cor- be related to the amount of alcohol consumed on day 0. Borrill relate inversely with aggression? A study using d-fenfluramine in et al. (1987) found that alcohol was associated with greater healthy subjects. Psychiatry Res. 69, 89–95. impairments of judgments of anger than of other emotions, Cooper, G., Allen, D., Scott, K., Oliver, J., Ditton, J., Smith, I., 2000. and both groups in the present study had consumed a large Hair analysis: self-reported use of ’speed’ and ’ecstasy’ compared amount of alcohol (a mean of over 10 units) before carrying with laboratory findings. J. Forensic Sci. 45, 400–406. Curran, H.V., Travill, R., 1997. Mood and cognitive effects of out the task on day 0. 3,4-methylenedioxymethamphetamine (MDMA, ‘ecstasy’): weekend It is tempting to speculate that heightened awareness of ’high’ followed by mid-week low. Addiction 92, 821–831. facial expression recognition in ecstasy users is a component Curran, H.V., 2000. Is MDMA (ecstasy) neurotoxic in humans? An of the acute subjective effect of increased empathy unique to overview of evidence and methodological problems in research. Neu- MDMA. However, there is no a priori reason why increased ropsychobiology 42, 34–41. Curran, H.V., Rees, H., Hoare, T., Hoshi, R., Bond, A.J., 2004. Empa- empathy should be reflected in increased accuracy to recog- thy and aggression: two faces of ecstasy? A study of interpretative nise only the single emotional expression of fear and not any bias and mood change in ecstasy users. Psychopharmacology 173, of the other basic emotions. 425–433. In summary, results from the present study indicate that Davis, M., Whalen, P.J., 2001. The amygdala: vigilance and emotion. an acute dose of MDMA may subtlety affect the recogni- Mol. Psychiatry 6, 13–34. Ekman, P., Friesen, W.V., 1976. Facial Action Coding System. Consulting tion of fearful facial expressions. These results compliment Psychologists Press, Palo Alto. previous findings that suggest that the recognition of fear is Ekman, P., 1992. An argument for basic emotions. Cognition Emotion 6, modulated by serotonin. Further, the differences in the pro- 169–200. cessing of fearful facial expressions occurred alongside in- Green, A., Cross, A., Goodwin, G., 1995. Review of pharmacology creases in aggression mid-week following weekend MDMA and clinical pharmacology of 3,4-methylenedioxymethamphetamine (MDMA or ‘ecstasy’). Pharmacology 119, 247–266. use. The fact that MDMA users are subtlety misinterpret- Harmer, C.J., Bhagwagar, Z., Cowen, P.J., Goodwin, G.M., 2002. Acute ing forms of social cues and are susceptible to negative administration of citalopram facilitates memory consolidation in feelings several days after taking the drug has implications healthy volunteers. Psychopharmacology 163, 106–110. 304 R. Hoshi et al. / Drug and Alcohol Dependence 76 (2004) 297–304

Harmer, C.J., Rodgers, R.D., Tunbridge, E., Cowen, P., Goodwin, G., Reneman, L., Booij, J., de Bruin, K., Reitsma, J., de Wolff, F., Gunnimg, 2003b. Tryptophan depletion decreases the recognition of fear in fe- W., den Heeten, G., Van de Brink, W., 2001. Effects of dose, sex, and male volunteers. Psychopharmacology 167, 411–417. long-term abstention from use on toxic effects of MDMA (ecstasy) Harmer, C.J., Bhagwagar, Z., Perrett, D.I., Vollm, B.A., Cowen, P., Good- on brain serotonin neurons. Lancet 358, 1864–1869. win, G., 2003a. Acute SSRI administration affects the processing Solowij, N., Hall, W., Lee, N., 1992. Recreational MDMA use in Sidney: of social cues in healthy volunteers. Neuropsychopharmacology 28, a profile of ecstasy users and their experiences with the drug. Brit. J. 148–152. Addict. 87, 1161–1172. Harris, D.S., Baggott, M., Mendelson, J., Jones, R.T., 2002. Subjec- Sprengelmeyer, R., Young, A.W., Schroeder, U., Grossenbacher, P.G., tive and hormonal effects of 3,4-methylenedioxymethamphetamine Federlein, J., But, T., Pruzuntek, H., 1999. Knowing no fear. Proc. (MDMA) in humans. Psychopharmacology 162, 396–405. Roy. Soc. Lond. [Biol] 226, 2451–2456. LeDoux, J.E., 2000. Emotional circuits in the brain. Annu. Rev. Neurosci. Thayer, J.F., Johnsen, B.H., 2000. Sex differences in judgement of facial 23, 155–184. affect: a multivariate analysis of recognition errors. Scand. J. Psychol. Lee, T.M.C., Liu, H.-L., Hoosain, R., Liao, W.-T., Wu, C.-T., Yuen, 41, 243–246. K.S.L., Chan, C.C.H., Fox, P.T., Gao, J.-H., 2002. Gender differences Tse, W.S., Bond, A.J., 2002. Serotonergic intervention affects both so- in neural correlates of recognition of happy and sad faces in humans cial dominance and affiliate behaviour. Psychopharmacology 161, assessed by functional magnetic resonance imaging. Neurosci. Lett. 324–330. 333, 13–16. Verheyden, S., Hadfield, J., Calin, T., Curran, H.V., 2002. Sub-acute Liechti, M., Gamma, A., Vollenweider, F., 2001. Gender differences in effects of MDMA (3,4-methylenedioxymethamphetamine, “ecstasy”) the subjective effects of MDMA. Psychopharmacology 154, 161–168. on mood: evidence of gender differences. Psychopharmacology 161, Mas, M., Farre, M., de la Torre, R., Roset, P.N., Ortuno, J., Segura, 23–31. J., Cami, J., 1999. Cardiovascular and neuroendocrine effects and Verheyden, S., Henry, J., Curran, H.V., 2003. Acute, sub-acute and long- pharmacokinetics of 3,4-methylenedioxymethamphetamine in humans. term subjective consequences of ’ecstasy’ (MDMA) consumption in J. Pharmacol. Exp. Ther. 290, 136–145. 430 regular users. Hum. Psychopharmacol. 18, 507–517. Nelson, H.E., O’Connoll, A., 1978. Dementia: the estimation of pre- Vollenweider, F., Gamma, A., Liechti, M., Huber, T., 1998. Psycholog- morbid intelligence levels using the New Adult Reading Test. Cortex ical and cardiovascular effects and short-term sequelae of MDMA 14, 234–244. (ecstasy) in MDMA naive healthy volunteers. Neuropsychopharma- Nichols, D.E., 1986. Differences between the mechanisms of action of cology 19, 241–251. MDMA, MBDB and the classic hallucinogens. Identification of a new Whalen, P.J., Rauch, S.L., Etcoff, N.L., McInerney, S.C., Lee, M.B., therapeutic class: entactogens. J. Psychoactive Drugs 8, 305–313. Jenike, M.A., 1998. Masked presentations of emotional facial ex- Parrott, A., Lasky, J., 1998. Ecstasy (MDMA) effects upon mood and cog- pressions modulate amygdala activity without explicit knowledge. J. nition: before, during and after a Saturday night dance. Psychophar- Neurosci. 18, 411–418. macology 139, 261–268. Young, A., Perrett, D., Calder, A., Sprengelmeyer, R., Ekman, P., 2002. Rapcsak, S.Z., Galper, S.R., Comer, J.F., Reminger, S.L., Nielsen, L., Facial Expressions of Emotion: Stimuli and Tests (FEEST). Thames Kasznaik, A.W., Verfaellie, M., Languna, J.F., Labiner, D.M., Co- Valley Test Company, UK. hen, R.A., 2000. Fear recognition deficits after focal brain damage: a Zigmond, A.S., Snaith, R.P., 1983. The hospital anxiety and depression cautionary note. Neurology 54, 575–581. scale. Acta Psychiatry Scand. 67, 361–370.