Psychopharmacology (2009) 205:85–91 DOI 10.1007/s00213-009-1517-4

ORIGINAL INVESTIGATION

Acute administration of the cannabinoid CB1 antagonist rimonabant impairs positive affective memory in healthy volunteers Jamie Horder & Philip J. Cowen & Martina Di Simplicio & Michael Browning & Catherine J. Harmer

Received: 11 August 2008 / Accepted: 11 March 2009 / Published online: 1 April 2009 # Springer-Verlag 2009

Abstract Background Emotional processing measures are sensitive to acute administration of clinically useful antidepressant drugs. We wished to test the hypothesis that these models would also be able to detect agents likely to cause depression as an adverse effect. The anti-obesity drug and cannabinoid type 1 receptor antagonist, rimonabant, is associated with significant rates of depression and anxiety in clinical use. Materials and methods Thirty healthy adult volunteers were randomly assigned to receive a single dose of rimonabant (20 mg) or lactose placebo in a double-blind, betweengroups design. Three hours after medication administration, subjects undertook an emotional processing test battery including facial emotion recognition, emotional word attentional dot probe, self-relevant word classification, emotional and declarative memory and the emotion-potentiated acoustic startle response. Subjective state was assessed via selfreport measures. Results A single dose of rimonabant did not alter subjective mood. However, rimonabant selectively reduced incidental recall of positive self-relevant adjectives, an effect contrary to that seen following the administration of antidepressants. There were no effects of rimonabant on the other measures of emotional processing. Conclusions These results suggest that a single dose of rimonabant decreases positive emotional memory in the absence of changes in subjective state. Further studies are required to examine whether rimonabant might produce a J. Horder : P. J. Cowen : M. Di Simplicio : M. Browning : C. J. Harmer (*) University Department of Psychiatry, Warneford Hospital, University of Oxford, Oxford, UK e-mail: [email protected]

wider range of negative emotional biases with repeated treatment. Keywords Rimonabant . CB1 antagonists . Healthy volunteers . Facial expression recognition . Cannabinoids . Depression

Introduction Negative biases in information processing are believed to play a key role in the development and maintenance of clinical depression. For example, depressed patients are more likely to retrieve negative self-relevant information in both explicit (i.e. free recall) and implicit (i.e. primed lexical decision) memory paradigms (Bradley et al. 1995). Also, depressed patients are more likely to perceive ambiguous facial expressions as negative compared to controls (Bouhuys et al. 1999; Gur et al. 1992), and this tendency predicts subsequent relapse (Bouhuys et al. 1999). These processing biases are believed to play a key role in the persistence of the depressed state, as increased accessibility of negative perceptions and memories maintains and exaggerates the depressed mood while being fuelled by the mood state, leading to a self-perpetuating cycle. We have hypothesised that antidepressant drugs directly modulate emotional information processing biases important in depression and that these effects are instrumental in producing subsequent changes in mood. Consistent with this theory, single doses of antidepressants have been found to facilitate the processing of positive emotional information in healthy volunteers in the absence of mood effects. For example, acute administration of the noradrenaline reuptake inhibitor (SNRI), reboxetine, improved the recogni-

86

tion of happy facial expressions, speeded emotional categorisation to positive adjectives and enhanced emotional memory for positive self-referent information (Harmer et al. 2003).These results suggest that changes in emotional processing may be a more sensitive measure of the early effects of antidepressant agents than are subjective measures of mood, which conventionally are thought to take a number of weeks to show clinically important improvement. By implication, such models might also be useful in the detection of agents which can produce symptoms of depression. Rimonabant (SR141716), an antagonist and possible inverse agonist (Pertwee 2005) at cannabinoid type 1 (CB1) receptors, was licenced in the European Union for the promotion of weight loss in clinically obese patients in 2006. However, rimonabant treatment is associated with an increased risk of both depression and anxiety (Christensen et al. 2007). In June 2007, a US Federal Drug Administration Advisory Panel recommended against approving rimonabant for use in the USA, despite its demonstrated efficacy in producing weight loss, due to concerns over the incidence of psychiatric symptoms in rimonabant users, in particular depression, suicidality and anxiety. In October 2008, rimonabant was withdrawn from use in Europe for the same reason following advice from the EU European Medicines Evaluation Agency (EMEA) regulatory agency (EMEA 2008) These serious and problematic psychiatric adverse effects were not predicted by commonly used animal psychopharmacology models, since CB1 antagonists, including rimonabant, have shown effects similar to those of antidepressants in some models such as the rodent forced swim (Porsolt) test (see e.g. Griebel et al. 2005). In the present study, we aimed to assess whether acute administration of rimonabant induces biases in emotional processing in a battery of psychological tasks. In particular, we wished to discover whether there was any evidence of negative emotional biases consistent with its side effect profile. Such effects are usually only identified after extensive use of these drugs by patients. The early detection of such effects could therefore have significant implications for public health by helping to restrict the exposure of vulnerable patients to potentially harmful compounds.

Materials and methods Participants and design We recruited 30 participants (16 female), with mean age 24.1 years (SD 5.4 years, range 18 to 40 years) who provided written informed consent for the study, which was approved by the local research ethics committee. Participants were not currently using psychotropic medication, were physically healthy as assessed by physical examina-

Psychopharmacology (2009) 205:85–91

tion and reported medical history and were free of current or past Axis I disorder on the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders-IV. Because of potential interactions between rimonabant and cannabinoids such as those in marijuana, participants were required to provide a negative urine drug screen for marijuana (SureScreen Diagnostics) on both the day of testing and the inclusion visit. For female subjects, testing was not carried out in the 7 days before the onset of menses. Participants were randomised to either rimonabant 20 mg (Acomplia, Sanofi-Aventis) or a lactose placebo, which were presented in identical gelatine capsules in a double-blind manner. From previous experience, we believed that some of the tasks used in this study were not suitable for repeated testing, owing to possible practice effects and because some of the tasks involve un-prompted responses (e.g. incidental memory). Hence, a between-subjects design was adopted. Subjects arrived at the laboratory at approximately 9:00 A.M. following an overnight fast. After a light breakfast, the treatment was administered with water at approximately 9:30 A.M. Psychological testing started (following a light lunch) 3 h later. Subjective ratings Before the day of testing, participants completed the Eysenck Personality Questionnaire (EPQ). On the day of testing, participants completed the Beck Depression Inventory (BDI), the Spielberger State-Trait Anxiety Inventory (STAI), the Befindlichkeits Scale (BFS) and visual analogue scales (VAS) for the following words: nausea, dizziness, hunger, anxiety, happiness, sadness, and alertness. VAS ratings were repeated at 60, 120 and 180 min post-treatment administration as well as final assessment at the end of the session (which was approximately 350 min after treatment); there were therefore five VAS time points. The BFS and the State scale of the STAI were given again at 180 min after treatment. Psychological tasks Rey Auditory Verbal Learning Task Declarative verbal memory was measured with the Rey Auditory Verbal Learning Test (AVLT; Rey 1964). Briefly, subjects were tested on immediate free recall of items on a 15-item word list over five repetitions (acquisition phase), followed by recall for a second 15-item word list, and then recall of the items on the first list. This was then followed by delayed recall and recognition of the same items approximately 20 min later. Word categorization and memory Sixty personality characteristic words selected to be extremely disagreeable (e.g. domineering, untidy, hostile) or agreeable (cheerful, honest, optimistic) (taken from Anderson 1968) were presented on

Psychopharmacology (2009) 205:85–91

the computer screen for 500 ms. These words were matched in terms of word length, ratings of frequency and meaningfulness. Participants were asked to categorise these personality traits as likable or dislikable, as quickly and as accurately as possible. Specifically, they were asked to imagine whether they would be pleased or upset if they overheard someone else referring to them as possessing this characteristic, so that the judgement was in part self-referring. Classifications and reaction times for correct identifications were computed for this task. Approximately 15 min after completion of the categorization task, participants were asked to recall and write down as many of the personality trait words as possible, and they were given 2 min to do this. This task therefore allowed the assessment of incidental memory for positive and negative characteristics. Recognition memory was then assessed by asking volunteers to respond with a ‘Yes’ or ‘No’ to each item on a list containing the 60 targets plus 60 matched distractors (30 positive, 30 negative). A similar categorization task was carried out as a control for non-specific effects on speed and incidental memory. This task also involved the presentation of an attribute word on the computer screen for 500 ms, but participants were asked to indicate whether the characteristic would be an advantage (30 words, e.g. strong) or disadvantage (30 words, e.g. slow) for a predatory animal. Words were thus matched in terms of imageability and frequency, but the judgement was not self-referring. Recall and recognition of these animal-related words were also assessed approximately 15 min after the categorization task using the same parameters as the emotional categorization task. The order of testing was: emotional word categorization, animal word categorization, emotional recall, emotional recognition, animal recall, and animal recognition. Dot-probe task Two types of emotional words were used in this task: 60 social-threatening negative words and 60 positive words. Each emotional word was paired with a neutral word that began with the same letter and was matched for length. There were therefore 60 socially threatening–neutral pairs and 60 positive–neutral pairs. In addition, there were 60 neutral– neutral pairs. Each trial started with a fixation cross for 500 ms, followed by a word pair. On each trial, one of the words appeared above and the other below the central fixation position. The emotional words appeared at the top and bottom location with equal frequency. In the unmasked condition, the word pair was presented for 500 ms and then a probe appeared in the location of one of the preceding words. The probe was either one or two stars, and participants were asked to press one of two labelled buttons to indicate the number of stars present on the screen; their response terminated the probe display. Participants were

87

asked to respond as quickly and as accurately as possible. The probe appeared at the top and the bottom of the screen with equal frequency. Thus, on half of the emotion–neutral word pair trials, the probe appeared in the same position as the emotional word, and on the other half, the probe appeared in the same position as the neutral word. The sequence of events was the same in the masked condition, except the duration of the word pair was 14 ms and the display of the word pair was immediately followed by a mask which was displayed for 186 ms. The mask was constructed from digits, letters and other non-letter symbols (for example @B%2#) and was matched for word position and length. There were 360 trials in total (masked—60 positive–neutral, 60 negative–neutral, 60 neutral–neutral; unmasked—60 positive–neutral, 60 negative– neutral, 60 neutral–neutral) and masked and unmasked trials were presented in a random order. Mean reaction time and accuracy scores were recorded. Reaction times which were at more than two standard deviations from individual means were considered as outliers and discarded. To simplify these results, attentional vigilance scores were calculated for each participant by subtracting the reaction time from trials when probes appeared in the same position as the emotional word (congruent trials) from trials when probes appeared in the opposite position to the emotional word (incongruent trials). Facial emotion recognition task The facial expression recognition task featured six basic emotions (happiness, surprise, sadness, fear, anger and disgust) taken from the Pictures of Affect Series (Ekman and Friesen 1976), which had been morphed between each prototype and neutral (Young et al. 1997). Briefly, this procedure involved taking a variable percentage of the shape and texture differences between the two standard images 0% (neutral) and 100% (full emotion) in 10% steps. Four examples of each emotion at each intensity were given (total of ten individuals). Each face was also given in a neutral expression, giving a total of 250 stimuli presentations. The facial stimuli were presented on a computer screen (random order) for 500 ms and replaced by a blank screen. Volunteers made their responses by pressing a labelled key on the keyboard. Participants were instructed to classify each face as being one of either angry, disgusted, fearful, happy, sad, surprised or neutral, as quickly and as accurately as possible. Accuracy and reaction time were measured in this task. Emotion-modulated startle response Stimuli Sixty-three pictures of three categories (pleasant, unpleasant, neutral) were taken from the International Affective Picture System (gender-specified; Larson et al. 2000). Each picture was presented for 13 s (mean inter-trial

88

interval=13 s) on a computer screen. The pictures were presented in three blocks in a fixed order such that no two of the same category would appear successively. Procedure and recording The eye-blink component of the startle response was recorded from the orbicularis oculi using electromyography (EMG startle response system, San Diego Instruments, San Diego, CA, USA). Acoustic probes were 50-ms, 95-dB bursts of white noise with a nearly instantaneous rise time (generated through the noise generator and amplifier of the EMG startle response system) and were delivered binaurally through headphones at 1.5, 4.5 or 7.5 s following picture onset. To minimise expectation, startle probes were skipped from two trials per valence per block, and three probes were given within the inter-trial interval. A practice session presenting nine neutral pictures and startle probes was used in the beginning to habituate participants to the startle probes. EMG signals were filtered (low cut-off, 0.5 Hz; high cutoff, 100 Hz) and rectified. Eye-blink reflex magnitudes in microvolts were calculated by subtracting the amount of integrated EMG at reflex onset from the first peak amplitude of integrated EMG between 20 and 120 ms following probe onset. Trials with no traceable eye-blink reflex were assigned a magnitude of zero and included in the analysis. Eye-blink reflexes with an excessively noisy baseline (within 20 ms after the probe) were rejected. Of the 30 volunteers, seven volunteers were not included in the analysis because of electrode interference (N=4) or because they displayed fewer than 25% satisfactory blink responses in the paradigm (N=3). This task provides a measure of the relative acoustic startle response during unpleasant, pleasant and neutral pictorial stimuli presentation. The enhanced startle response during a state of fear in rodents is used to assess anxiolytic properties of drugs (Davis et al. 1993). Given the anxiogenic effects of CB1 antagonism in most animal studies, it was predicted that rimonabant would specifically enhance the effects of the negative pictures on acoustic startle (i.e. the heightened acoustic startle during viewing of the unpleasant relative to the neutral pictures). Statistical analysis The demographic characteristics of the two groups were compared using independent samples t tests. Subjective VAS scores were subjected to repeated-measures analysis of variance (ANOVA) with drug status as a between-subjects factor and time from baseline as a within-subject factor (four change values per item). BFS and STAI data were analysed in terms of change from baseline using independent values t test. Data from the facial emotion recognition task, the dot

Psychopharmacology (2009) 205:85–91

probe task, the word categorization and memory task and the emotion-potentiated acoustic startle paradigm were assessed using repeated-measures ANOVA with drug as a between-subjects factor and stimulus emotional valence as a within-subject factor. For the dot-probe, masking was also a within-subject factor. Statistically significant interactions were followed up with simple main effect analyses. Data from each task were analysed separately without correcting for multiple comparisons across tasks.

Results Demographic characteristics The treatment groups were well matched in terms of gender, age, verbal IQ as estimated using scores on the National Adult Reading test (NART), baseline scores on the BDI, BFS, EPQ N-scale (Neuroticism), state measure of the STAI and trait measure of the STAI (see Table 1; all p>0.5, two-tailed). Subjective state ratings There was no between-group difference in the change in subjective state ratings following rimonabant versus placebo for the VAS, BFS and STAIS (all p>0.1). Emotional processing tasks Word categorization ANOVA revealed no significant group or group × valence effect for either accuracy or reaction time to classify self-

Table 1 Baseline and demographic measures in the two groups of volunteers Measure

Drug

N

Mean

p value

BDI

Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug

15 15 15 15 15 15 15 13 15 15 15 15 15 15

2.7 2.6 30.4 30.1 5.9 6.0 18.3 17.5 23.5 24.7 17.4 14.9 28.8 27.7

0.936

STAI-T EPQ-N NART (words incorrect) Age (years) BFS baseline STAI-S baseline

0.890 0.966 0.774 0.581 0.651 0.602

Psychopharmacology (2009) 205:85–91

relevant words as positive vs. negative (accuracy—group × valence F(1,27)=0.012, p=0.912; main effect of group F(1,27)=3.446, p=0.074; reaction time—group × valence F(1,27)=1.646, p=0.210; main effect of group F(1,27)= 1.289, p=0.266). There was also no effect on the categorisation of non self-relevant words (accuracy—group × valence F(1,28)<0.001, p=1.00; main effect of group F (1,28)=0.027, p=0.871; reaction time—group × valence F (1,28)=1.322, p=0.260; main effect of group F(1,28)= 0.004, p=0.953).

89

Word recognition There were no significant between-group differences in the accuracy or latency of recognition of previously seen selfrelevant or non-self-relevant words (for self-relevant words, accuracy group × valence ANOVA (n=30) p=0.626; reaction time p=0.703; main effect of group, accuracy p= 0.085; reaction time p=0.558; for non-self-relevant words, accuracy group × valence ANOVA (n=30) p=0.269; reaction time p=0.467; main effect of group, accuracy p= 0.145; reaction time p=0.590; see Table 2).

Word recall Words dot probe Rimonabant significantly impaired the recall of positive items in the self-referent emotional memory task (group × valence F(1,28)=4.204, p=0.05; t test on positive words recalled, p=0.028, Fig. 1). By contrast, there was no effect of the drug in the non-self-relevant words control task (ANOVA F(1,27)=2.673, p=0.114). There was no main effect of the drug in either task (self-relevant, F(1,28)=2.701, p=0.111; non-self-relevant, F(1,27)=0.347, p=0.561). In order to verify that these effects were not the result of gender-specific effects, we performed the same analyses using gender as an additional within-subject factor. In the case of the emotional words, this confirmed a significant drug effect (group × valence F(1,26)=4.309, p=0.048) and no interaction with gender (group × valence × gender F (1,26)=0.81, p=0.376). In the case of the animal words, this confirmed that there were no significant effects or interactions (all p>0.18).

ANOVA revealed no significant group effects or interactions with group (all p>0.195). Separate ANOVAs for the masked and for the unmasked condition showed no significant effects either (masked, all p>0.624, unmasked, all p>0.141). Facial emotion recognition Considering accuracy to classify faces as each of the six basic emotions, ANOVA revealed a highly significant main effect of emotion (F(5,140)=8.196, p<0.001) but no emotion × group interaction (F(5,140)=0.232, p=0.948) or group main effect (F(1,28)=0.029, p=0.867, Fig. 2). A series of independent samples t tests revealed no significant group differences on accuracy to recognise any emotion, even before correcting for multiple comparisons (all p> 0.527 uncorrected). For response latency, ANOVA (n=30) revealed a main effect of emotion (p<0.001), with no main effect of group (p =0.793) and no emotion × group interaction (p>0.05). Rey AVLT A repeated-measures ANOVA on words correctly recalled over the five acquisition trials of the AVLT (n=30) revealed a highly significant effect of time (F(4,112)=89.077, p< 0.001) indicating successful verbal learning but no group × time interaction (F(4,112)=0.367 p=0.783) nor a group main effect (F(1,28)=1.648, p=0.210). A series of t tests on each of the 11 separate outcome variables confirmed no significant group differences (all p≥0.185 uncorrected). Emotion-modulated acoustic startle

Fig. 1 Incidental verbal recall task. Values are mean number of words correctly recalled ±1 SEM. Asterisks represent the statistical significance of comparisons between groups; *p<.05

ANOVA, with the outcome measure being the absolute startle voltage recorded, with emotional valence of images as a within-subject factor, revealed an effect of emotion (F(2,44)= 4.676, p=0.014) but no main effect of group (F(1,22)=0.018, p=0.886) or emotion × group interaction (F(2,44)=0.954,

90

Psychopharmacology (2009) 205:85–91

Table 2 Emotional categorisation and recognition task results Task

Drug

N

Mean

Categorization accuracy—positive self-relevant

Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug Placebo Drug

15 14 15 14 15 15 15 15 15 15 15 15 15 15 15 15

29.40 28.43 28.60 27.71 28.60 28.53 28.67 28.60 25.1333 22.6000 20.6667 17.3333 25.7333 24.4000 20.4667 17.4667

Categorization accuracy—negative self-relevant Categorization accuracy—positive non-self-relevant Categorization accuracy—negative non-self-relevant Recognition accuracy—positive self-relevant Recognition accuracy—negative self-relevant Recognition accuracy—positive non-self-relevant Recognition accuracy—negative non-self-relevant

p=0.393). Startle amplitude was modulated by the valence of the picture as expected (i.e. aversive pictures potentiated the response), but rimonabant did not influence this effect, nor had it any effects on startle amplitude overall.

Discussion Our results indicate that a single oral dose of rimonabant, 20 mg, selectively reduced incidental recall of positive selfrelevant words in the absence of any effects on subjective state. This effect is consistent with the observation that

Fig. 2 Facial emotion recognition task. Values are mean accuracy levels for each emotion following treatment with rimonabant or placebo ±1 SEM

Std. deviation 0.828 1.989 1.404 2.268 1.121 1.885 0.816 1.454 2.55976 5.91366 4.04734 6.56470 1.94447 4.89606 3.87052 6.03403

p 0.108 0.213 0.907 0.878 0.144 0.105 0.340 0.116

rimonabant can produce symptoms of depression in clinical practice and suggests that performance on self-referent emotional memory tasks may be a selective marker of the depressogenic potential of novel therapeutic agents. We observed no effect of rimonabant upon declarative verbal memory performance as assessed using the Rey AVLT task nor upon incidental memory for classified words except in the case of positive self-referential words. The effect seen here was therefore selective to words with a personal, positive relevance and not a general memory-impairing effect. A similar reduction in memory for positive affective stimuli has also been reported in depression and has been suggested to play a significant role in the maintenance of depression as increased accessibility of negative perceptions and memories maintains and exaggerates the depressed mood (Segal et al. 2006). This same emotional memory task is also affected by acute and repeated antidepressant drug administration in healthy volunteers: a single dose of reboxetine or duloxetine increased recall of positive emotional items in this task (Harmer et al. 2008; Harmer et al. 2006) contrary to the effect reported here. It is noteworthy that, in the current study and previously, we have observed effects of psychotropic drugs on emotional processing in the absence of any changes in subjective state. Indeed, we have hypothesised that these effects on emotional processing may be a key mechanism by which antidepressants lead to enhanced mood over time. Similarly, it is possible that drugs such as rimonabant which are associated with increased rates of depression and anxiety in clinical use may do so by reducing positive affective memories which may over time lead to reduced levels of positive self-associations and impaired mood. The effect of rimonabant in the current study was, however, quite small and very selective, in that there was

Psychopharmacology (2009) 205:85–91

no effect upon other measures of emotional processing which have been associated with both antidepressant drug administration and the state of depression (Bouhuys et al. 1999; Gur et al. 1992; Harmer et al. 2003) such as facial expression recognition accuracy. Further studies are therefore required to assess whether a broader range of changes would be seen with repeated drug administration and higher plasma levels of rimonabant or if there is any modulation of these effects by gender. In addition, the effects of this drug may require interaction with vulnerability to depression and anxiety and may be expected to be greater in those with a history of or vulnerability to depressive illness. The current results also contrast with findings from preclinical animal studies which have shown both antidepressant-like effects (e.g. Griebel et al. 2005) and enhanced memory processes (Lichtman 2000; Shiflett et al. 2004; Terranova et al. 1996; Wolff and Leander 2003) following acute rimonabant administration. These differences may relate to differences in tasks applied. In particular, human studies allow greater exploration of verbal affective processing which cannot be tapped in animal studies. Animal studies are also able to use a range of doses which may allow effects on memory to be more effectively detected. Further exploration of these effects is needed in analogue tasks which can be studied in man and rodents such as spatial working memory with longer-term administration. In conclusion, our results suggest that a single dose of rimonabant has an effect upon an aspect of emotional processing that is opposite to that seen with acute antidepressant treatment and that is therefore consistent with its liability to cause depressive mood changes in some patients. Further studies using longer-term rimonabant administration, perhaps in more vulnerable individuals, will be needed to understand the mechanisms by which rimonabant increases the risk of anxiety and depression during clinical use. It is also possible that early changes in emotional processing following initiation of rimonabant treatment could be helpful in detecting patients particularly likely to experience adverse mood effects during treatment. Acknowledgements Rena Hockney.

Thanks to Matthew Taylor, Danilo Arnone and

Funding and conflict of interest This study was supported by the MRC. Catherine Harmer serves on the advisory panel for p1vital and has acted as a paid consultant for Lundbeck, Merck, Sharpe and Dohme and p1vital. Philip Cowen has acted as a paid consultant for Eli Lilly, Lundbeck and Servier.

References Anderson NH (1968) Likableness ratings of 555 personality-trait words. J Pers Soc Psychol 9:272–279

91 Bouhuys AL, Geerts E, Gordijn MC (1999) Depressed patients' perceptions of facial emotions in depressed and remitted states are associated with relapse: a longitudinal study. J Nerv Ment Dis 187:595–602 Bradley BP, Mogg K, Williams R (1995) Implicit and explicit memory for emotion-congruent information in clinical depression and anxiety. Behav Res Ther 33:755–770 Christensen R, Kristensen PK, Bartels EM, Bliddal H, Astrup A (2007) Efficacy and safety of the weight-loss drug rimonabant: a meta-analysis of randomised trials. Lancet 370:1706–1713 Davis M, Falls WA, Campeau S, Kim M (1993) Fear-potentiated startle: a neural and pharmacological analysis. Behav Brain Res 58:175–198 Ekman P, Friesen WV (1976) Pictures of facial affect. Consulting Psychologists Press Palo Alto, CA EMEA (2008) PRESS RELEASE: The European Medicines Agency recommends suspension of the marketing authorisation of Acomplia. http://www.emea.europa.eu/humandocs/PDFs/EPAR/ acomplia/53777708en.pdf European Medicines Agency Press Office Griebel G, Stemmelin J, Scatton B (2005) Effects of the cannabinoid CB1 receptor antagonist rimonabant in models of emotional reactivity in rodents. Biol Psychiatry 57:261–267 Gur RC, Erwin RJ, Gur RE, Zwil AS, Heimberg C, Kraemer HC (1992) Facial emotion discrimination: II. Behavioral findings in depression. Psychiatry Res 42:241–251 Harmer CJ, Hill SA, Taylor MJ, Cowen PJ, Goodwin GM (2003) Toward a neuropsychological theory of antidepressant drug action: increase in positive emotional bias after potentiation of norepinephrine activity. Am J Psychiatry 160:990–992 Harmer CJ, Mackay CE, Reid CB, Cowen PJ, Goodwin GM (2006) Antidepressant drug treatment modifies the neural processing of nonconscious threat cues. Biol Psychiatry 59:816–820 Harmer CJ, Heinzen J, O'Sullivan U, Ayres RA, Cowen PJ (2008) Dissociable effects of acute antidepressant drug administration on subjective and emotional processing measures in healthy volunteers. Psychopharmacology (Berl) 199:495–502 Larson CL, Ruffalo D, Nietert JY, Davidson RJ (2000) Temporal stability of the emotion-modulated startle response. Psychophysiology 37:92–101 Lichtman AH (2000) SR 141716A enhances spatial memory as assessed in a radial-arm maze task in rats. Eur J Pharmacol 404:175–179 Pertwee RG (2005) Inverse agonism and neutral antagonism at cannabinoid CB1 receptors. Life Sci 76:1307–1324 Rey A (1964) L’examen de clinique en psychologie. Paris, Presses Universitaires de France Segal ZV, Kennedy S, Gemar M, Hood K, Pedersen R, Buis T (2006) Cognitive reactivity to sad mood provocation and the prediction of depressive relapse. Arch Gen Psychiatry 63:749–755 Shiflett MW, Rankin AZ, Tomaszycki ML, DeVoogd TJ (2004) Cannabinoid inhibition improves memory in food-storing birds, but with a cost. Proc Biol Sci 271:2043–2048 Terranova JP, Storme JJ, Lafon N, Perio A, Rinaldi-Carmona M, Le Fur G, Soubrie P (1996) Improvement of memory in rodents by the selective CB1 cannabinoid receptor antagonist, SR 141716. Psychopharmacology (Berl) 126:165–172 Wolff MC, Leander JD (2003) SR141716A, a cannabinoid CB1 receptor antagonist, improves memory in a delayed radial maze task. Eur J Pharmacol 477:213–217 Young AW, Rowland D, Calder AJ, Etcoff NL, Seth A, Perrett DI (1997) Facial expression megamix: tests of dimensional and category accounts of emotion recognition. Cognition 63:271– 313