Participants

All experimental procedures were endorsed by the Bethlem Royal and Maudsley research ethics committee and the ethics committee at Dresden University Medical Centre. All participants gave written informed consent to the scientific use of their data and were reimbursed for their participation. The study was conducted in compliance with the Helsinki declaration. ¹⁰ The study included a sample of 9 individuals (5 men and 4 women, mean age 36.11 years (s.e.m.=2.34); educational level 2.22 (s.e.m.=0.14), where 2 represents undergraduate level), with a primary diagnosis of depersonalisation disorder from the Maudsley Hospital outpatient department. At the time of this investigation the patients were being treated in a specialised clinic for this condition. All patients were re-examined and confirmed with diagnosis of depersonalisation according to DSM–IV criteria by a psychiatrist not involved in the study. Patients with a confirmed diagnosis were separately invited to participate in the study by the experimenter, who was masked to all medical records (E.L.). The clinical cut-off level of a score above 70 on the Cambridge Depersonalization Scale (CDS) for depersonalisation disorder was exceeded for all patients (175.77, s.e.m.=12.31). ^{Reference Sierra and Berrios11} Three people with depersonalisation disorder received minimum effective doses of one of three different substances: including selective serotonin reuptake inhibitors (SSRIs) and neuroleptics (paroxetine, fluoxetine, olanzapine). A normal control group of 12 participants (7 men and 5 women, mean age 27.25 years (s.e.m.=1.95); educational level 2.58 (s.e.m.=2.02)) was also included. These participants were chosen to match the sample characteristics of the depersonalisation disorder group, specifically with respect to global intellectual functioning and socio-demographic features. A trend towards significance in age differences seemed tolerable, because none of the study variables showed any association with participants' age.

Self-report questionnaire data

All participants completed self-report forms before being introduced to the experimental protocol inside the scanner. Right-handedness was verified with the Edinburgh Handedness Inventory. ^{Reference Oldfield12} Further to the CDS clinical cut-off measure for depersonalisation, ^{Reference Sierra and Berrios11} clinical dimensions potentially relevant for depersonalisation disorder were assessed on the day of scanning, using the Dissociative Experience Scale, ^{Reference Bernstein and Putnam13} the Screening for Somatoform Disorders, ^{Reference Rief, Hiller and Heuser14} the Toronto Alexithymia Scale (20-item version), ^{Reference Taylor, Bagby, Ryan, Parker, Doody and Keefe15,Reference Taylor, Bagby, Ryan and Parker16} the Frankfurt Body Concept Scales, ^{Reference Deusinger17} the Beck Depression Inventory, ^{Reference Beck, Ward, Mendelson, Mock and Erbaugh18} and the State–Trait Anxiety Inventory. ^{Reference Spielberger19} Discriminative cut-off levels for depersonalisation disorder have not been established for these instruments; however, they served as additional measures of symptom severity.

Implicit facial expression neuroimaging tasks

The participants completed two 6 min experiments employing event-related functional magnetic resonance imaging (fMRI). In each experiment, participants were presented with ten different facial identities, each expressing twice 50% and 100% intensities of one emotion (either happiness or sadness) in addition to a neutral expression (0%) (60 facial stimuli, 12 non-facial stimuli). Each facial stimulus was presented for 2 s. During the interstimulus interval, the duration of which varied from 3 s to 8 s according to a Poisson distribution at an average interval of λ=4.9 s, participants viewed a fixation cross, as described elsewhere. ^{Reference Surguladze, Brammer, Young, Andrew, Travis, Williams and Phillips20,Reference Surguladze, Brammer, Keedwell, Giampietro, Young, Travis, Williams and Phillips21} Further details of the fMRI paradigm are presented in a data supplement to the online version of this paper.

Psychophysiological recording

Derivations of electrodermal activity related to the task were made online during neuroimaging data acquisition. The method of simultaneous fMRI and psychophysiology data acquisition and analysis used here has been described in detail elsewhere. ^{Reference Lemche, Giampietro, Surguladze, Amaro, Andrew, Williams, Brammer, Lawrence, Maier, Russell, Simmons, Ecker, Joraschky and Phillips22} Applying criteria of 0.01 μS, electrodermal activity was analysed in an event-related manner for each of the three different emotion expression intensities in each participant using the software program SC-ANALYZE (Neuroimaging Research Group, Institute of Psychiatry, London, UK). Latency windows of 1.2–3.3 s post-stimulus onset were evaluated to ensure that electrodermal activity was not contaminated by non-specific skin conductance response (SCR) discharges (increasing skin conductance level, (SCL)), such as those to the following stimulus. The following electrodermal activity variables were submitted to statistical analysis: SCR rates, SCL latency, SCL amplitude height, mean SCL, minimum SCL and maximum SCL in each time window. Following standard procedures, two variables were computed, ΔSCL and relative SCL (rSCL), expressing individual spans between minima and maxima within each condition, and the relative means normalised to these. ^{Reference Lemche, Giampietro, Surguladze, Amaro, Andrew, Williams, Brammer, Lawrence, Maier, Russell, Simmons, Ecker, Joraschky and Phillips22}

Image acquisition and analysis

Gradient echo echoplanar imaging (EPI) data were acquired on a neurovascular GE Signa 1.5 T system (General Electric, Milwaukee, Wisconsin, USA), equipped with 40 m/mT high-speed gradients, at the Maudsley Hospital, London. A quadrature birdcage head-coil was used for radiofrequency transmission and reception. For each of the two tasks, 180 T₂*-weighted images were recorded over 6 min at each of 16 near-axial non-contiguous 7 mm thick planes parallel to the anterior–posterior commissural (AC–PC) line: echo time (TE) 40 ms, repetition time (TR) 2000 ms, in-plane resolution 3.44 mm, interslice gap 0.7 mm, flip angle (FA) α=70°, matrix 64×64, field of view (FOV) 25 cm providing whole brain coverage. During the same session a high-resolution anatomical data-set was acquired with an EPI pulse sequence. The structural images were acquired at 43 near-axial planes 3 mm thick parallel to the AC–PC line: TE 73 ms, inversion time (TI) 180 ms, TR 16 000 ms, in-plane resolution 1.72 mm, interslice gap 0.3 mm, matrix size 128×128, FOV 25 cm, FA α=90°. The high-resolution EPI data-set was later used to register the fMRI data-sets acquired from each individual in standard stereotaxic space. The program package XBAM for UNIX (www.brainmap.it) with mathematical control for signal-to-noise ratio was used to perform the analysis of fMRI data. A detailed description of the fMRI analysis method is presented as a data supplement to the online version of this paper.

Behavioural performance during the gender decision task

Judgement accuracies in the gender decision task for facial expressions were evaluated as the percentage of correct answers in each of the six categories (neutral, mild and intense, for both happiness and sadness expressions). Complete descriptive data for the two study groups are presented in a data supplement to the online version of this paper. Correct overall answers were 49.54% for the depersonalisation disorder group and 51.31% for the control group. These rates around chance reflect relative task difficulty and are in line with other studies using similar fast implicit facial paradigms. ^{Reference Surguladze, Brammer, Keedwell, Giampietro, Young, Travis, Williams and Phillips21,Reference Van der Veen, Evers, Deutz and Schmitt23,Reference Deeley, Daly, Surguladze, Tunstall, Mezey, Beer, Ambikapathy, Robertson, Giampietro, Brammer, Clarke, Dowsett, Fahy, Phillips and Murphy24} No systematic difference between the depersonalisation disorder group and the control group emerged for reaction times or response accuracies.

Psychometric evaluation

The descriptive values for the questionnaire data are listed in a data supplement to the online version of this paper. No significant difference in handedness or on any of the nine taxons of the Frankfurt Body Concept Scales was found between the two groups. Significant between-group differences were observed for the Dissociative Experience Scale, the Screening for Somatoform Disorders, the CDS, the Toronto Alexithymia Scale (except its taxon for external-concrete cognitive style), the Beck Depression Inventory and the Spielberger State–Trait Anxiety Inventory. In all of these dimensions, scores were higher for participants with depersonalisation disorder than for the control group. The significant group differences, however, do not address clinical cut-off levels for specific disorders, except for the CDS (see Method).

Skin conductance levels

All SCL variables exhibited between-condition stability and between-emotion discrimination at each level (data not shown). Descriptive data are presented in the data supplement to the online version of this paper for ΔSCL and rSCL. Electrodermal activity data were not confounded by recording times and dates, but gender and education exhibited significant interaction effects in preliminary analyses. Consequently, the contribution of these confounding variables was removed by treating them as covariates in analyses of covariance (ANCOVAs). Significant between-group differences were apparent for five of the six stimulus types, and a near-significant difference for 50% sadness. Delta SCL was the electrodermal activity variable showing between-group differences to neutral (happy context), neutral (sad context), 50% sadness and 50% happiness stimuli (data presented in a data supplement to the online version of this paper), whereas rSCL was the variable showing a significant between-group difference to 100% happiness and 100% sadness stimuli (P<0.05). Comparison of the error diagrams (presented in a data supplement to the online version of this paper) (Fig. 1) revealed that the depersonalisation disorder group had much larger variabilities in electrodermal activity to facial stimuli than the control group. Both ΔSCL and rSCL means were higher in the depersonalisation disorder group, with the sole exception of rSCL in the 100% sadness condition. These findings suggest that it is the magnitude of the span between individual minima and maxima that discriminates best between depersonalisation disorder and control groups at 0% and 50% intensity levels; at the 100% level, the mean corrected measure was most effective in discriminating between groups.

Polynomial trend analyses of neural response

Linear trend maps reflecting greater neural response with linear increase in emotion intensity of expressions from neutral through mild to intense emotion are shown in Fig. 1. Talairach coordinates for regions showing significant linear increases in response to expressions of increasing emotion intensity are listed in a data supplement to the online version of this paper. The trend map for happy expressions in the control group (Fig. 1(a)) showed activation in the left orbital gyrus (Brodmann area (BA) 11), left middle frontal gyrus (BA 9), left angular gyrus (BA 39), left posterior cingulate gyrus (BA 31) and right fusiform gyrus (BA 19). In depersonalisation disorder the trend map for happy expressions (Fig. 1(b)) showed activation in the right superior frontal gyrus (BA 8), left temporal pole (BA 38), posterior insula and bilateral fusiform gyrus (BA 19). The trend map for sad expressions in the control group (Fig.1(c)) showed activation in the bilateral middle frontal gyrus (BA 8), posterior inferior temporal cortex (BA 20/39), left supramarginal gyrus (BA 40) and bilateral middle occipital gyri (BA 19). The corresponding map in the depersonalisation disorder group (Fig.1(d)) showed activation in the right superior frontal gyrus (BA 9/45), posterior insula, left supramarginal gyrus (BA 40), bilateral middle occipital gyri (BA 19) and left fusiform gyrus (BA 37).

Trend comparison analyses of neural response

The trend comparison maps reflecting interaction effects, i.e. regions where polynomial trends in neural response to expressions of increasing intensity of happy or sad emotion differed significantly between groups, are shown in Fig. 2. Note that these can also include non-linear (i.e. quadratic) effects. The graphs in this figure illustrate these interaction effects by showing the percentage blood oxygen level dependent (BOLD) signal change extracted from each identified region in the trend comparison map for each separate expression intensity. Analyses of between-group differences for happiness trends (Fig. 2(a)) identified the right hypothalamus (anterior portion at paraventricular nucleus, superior to the hypophyseal peduncle) and for sadness (Fig. 2(b)) the right amygdala (centromedial nucleus close to the processus uncinatus) as main clusters. Repeated-measures analyses of variance (ANOVAs) conducted on extracted signal intensities to examine between-group differences in polynomial trends for each emotion trend map (happiness F _1,20=4.522, P=0.048; sadness F _1,20=7.808, P=0.005) confirmed the findings from the above whole-brain trend comparison maps. Between-group differences to emotional stimuli, based on logarithmised fMRI signal effect sizes from the above regions extracted at 6 s post-stimulus, were greatest at 100% intensity levels: happiness, t _18,20=–2.134 (95% CI 0.019 to −2.551), P=0.047; sadness, t _18,20=–2.103 (95% CI 0.009 to −2.224), P=0.050. The profile plots for these and the secondary clusters (not shown) showed that the depersonalisation disorder group exhibited decreases in fMRI signal to expressions of increasing intensity of emotion. The opposite pattern was evident for the control group.

BOLD time course examination for neutral, mild and intense expressions

To examine further neural responses in the main clusters in the trend comparison maps discriminating between control and depersonalisation disorder, time courses for each of the three expression intensity levels (neutral–mild–intense) were extracted from regions in which between-group effects were shown in the trend comparison maps. The averaged time series of percentage change in BOLD signal are plotted in Fig. 3. Time points represent units of 2 s post-stimulus (TR units; repetition time 2 s). To fulfil the criterion for a ‘peak’ in haemodynamic response, a positive or negative deflection in BOLD response had to exceed the preceding data-point by the standard deviation indicated by its error bar. The depersonalisation disorder group showed early positive or negative peak haemodynamic responses at TR=1, as indicated by error bars (representing 1 standard deviation). The control group showed initial peak haemodynamic responses at TR=2 or later. The average haemodynamic response is expected 4–6 s post-stimulus in non-visual heteromodal areas. ^{Reference Bandettini, Wong, Hinks, Tikovsky and Hyde25,Reference Aguirre, Zahran and D'Esposito26}

To investigate further the time-course pattern exhibited by the depersonalisation disorder group, we decided to examine haemodynamic responses in other regions. We correlated statistical maps of the subtraction contrast intense–mild expression intensity (thereby removing face-related activations, and preserving mid-to-high range emotion activation) for each of the emotions with the respective SCL measures that had discriminated between depersonalisation disorder and control (rSCL and ΔSCL). From the resulting equal number of regions in which significant positive correlations were shown (14 for each group), we extracted time series of haemodynamic responses. These regions included in both groups subcortical, ventral prefrontal cortical and visual processing neural regions, previously implicated in the response to emotional facial expressions. ^{Reference Surguladze, Brammer, Young, Andrew, Travis, Williams and Phillips20} The average BOLD response time courses for the three expression intensity levels for each emotion were plotted for all regions detected in whole-brain correlation images in each group. To fulfil the criterion for a peak in haemodynamic response, a positive or negative deflection in BOLD response had to exceed the preceding data-point by the standard deviation indicated by its error bar. There were 14 clusters for depersonalisation disorder, 13 of which had a peak at 2 s post-stimulus, and 1 at 4 s post-stimulus or later. In contrast, in the control group, one region of interest had a first peak at 2 s post-stimulus and 13 first peaks at 4 s post-stimulus or later. The difference between groups in number of early (TR=1) and late (TR=2 or later) peaks was significant (χ² _1,28=14.640, Fisher's exact test P<0.001). In a majority of the regions of interest in correlation maps for depersonalisation disorder, each of the higher emotion intensity expressions was associated with earlier (positive or negative) peaks in the time series compared with neutral expression. This supported the findings regarding timings of first peak in haemodynamic response for analysis of time series in the trend comparison maps (Fig. 3). Time series data for all evaluated regions of interest are given in a data supplement to the online version of this paper.

Fig. 1 Emotion-specific trend maps. Happiness and sadness intensity linear trend maps for the control and depersonalisation disorder groups (<0.14–0.16 error clusters over the entire brain). (a) Happiness, control group; (b) happiness, disorder group; (c) sadness, control group; (d) sadness, disorder group. Regions shown exhibit main effects for continual increases from neutral expression to 50% to 100% intensity of expression, relative to fixation cross baseline. Numbers below the slices indicate Talairach z coordinates. A colour version of this figure showing regions of activation can be found on the online version of this paper.

AMY, amygdala; BA, Brodmann area; CBM, cerebellum; FEF, frontal eye fields; FFG, fusiform gyrus; HIP, hippocampus; INS, insula; IPL, inferior parietal lobule; ITG, inferior temporal gyrus; MOG, middle occipital gyrus; MTG, middle temporal gyrus; OFC, orbitofrontal cortex; PCC, posterior cingulate cortex.

Limitations of the study

Among the limitations of this study is the small size of the sample, owing to the rarity of depersonalisation disorder as a primary or single diagnosis. Potential medication effects could not be partialled out statistically for three reasons: the small number of participants receiving medication (n=3, all with depersonalisation disorder); the low dosages of the medication; and the different medication sub-classes taken by these participants. It is emphasised, however, that the majority of individuals with depersonalisation disorder were unmedicated, and three received the lowest doses known to be effective. Additional group maps (not shown) for the unmedicated participants with depersonalisation disorder revealed highly similar cerebral activation patterns compared with the complete sample; we are thus able to rule out medication effects as the source of the presented results.

Directions for further research

Decreases in amplitude of response in neural regions implicated in emotion processing to emotional stimuli of increased, relative to those of decreased, emotional intensity, co-engagement of inhibitory prefrontal regions, together with early coupling between neural and autonomic responses to positive and negative emotional stimuli, may underlie the emotional blunting observed in depersonalisation disorder. It is likely that accelerated emotional appraisal of facial cues may lead to the subsequent downregulation of emotional experiences reported in depersonalisation disorder. Future studies measuring neural and autonomic responses to emotional stimuli in larger numbers of individuals with depersonalisation disorder will help elucidate the neural mechanisms underlying the emotional blunting in this disorder, and will help to increase understanding of the neural mechanisms underlying involuntary inhibition of emotional experiences per se.