Isr J Psychiatry - Vol. 55 - No 1 (2018) Lily Rothschild-Yakar et al. “Eating Me Up from Inside”: A Pilot Study of Mentalization of Self and Others and Emotion Regulation Strategies among Young Women with Eating Disorders

Lily Rothschild-Yakar, PhD,1 Merav Peled, MA,1 Adi Enoch-Levy, MD,2 Eitan Gur, MD,2 and Daniel Stein, MD2,3

1 Department of , University of Haifa, Haifa, Israel 2 Safra Children’s Hospital, Sheba Medical Center, Tel Hashomer, Israel 3 Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel

Introduction Abstract This study examines the dynamics of eating disorders Background: We examined the relationship between (EDs) in the framework of the mentalization-based model general ability of mentalization, the specific aspect proposed by Fonagy et al. and Fonagy and Target (1, 2), of affective mentalizing of self and others, emotion in order to improve our understanding of the way in regulation strategies, and eating disorder (ED) symptoms. which personality factors are related to the clinical picture Method: Twenty-five female adolescent and young of EDs. Specifically, we aim to explore the relationship adult inpatients with EDs, and 22 healthy subjects, between the general ability of mentalization, and the were administered a semi-structured interview - the specific component affective mentalization - regarding Reflective Function (RF) scale, self-rating scales self and others, and the strategies of emotion regulation assessing , emotion regulation, depression in patients with EDs compared to controls. and ED symptomatology, and a neurocognitive measure assessing . Mentalization and Reflective Functioning Fonagy and colleagues (1, 3) suggested a model of men- Results: Participants with EDs presented lower levels talization that provides a formulation for the normal of RF regarding the self and higher levels of alexithymia, development of modes of experience and psychopathol- using more emotional suppression and less cognitive ogy. This model integrates theories of the social-cognitive reappraisal than controls. Elevated levels of general concept of “theory of mind,” attachment, object-relations RF and self RF and attenuated alexithymia, along with and self-psychology. Mentalization refers to the capacity elevated cognitive reappraisal and attenuated emotional to reflect on and interpret one’s own behavior and that of suppression, were correlated with attenuated ED others based on intentional internal mental states, such as symptoms. Comorbidity with depressive symptoms beliefs, thoughts and emotions. The ability to mentalize predicted greater ED symptomatology. the experiences of self and others enables the individual Conclusions: High mentalization may serve as a coping to cope with external and internal stressors, to regulate mechanism to attenuate ED symptoms. affects and to form good interpersonal relationships (1, 3). A major factor facilitating children’s attainment of the ability to mentalize is based on the parents’ attuned mirroring of their affective states, thus providing a secure base for the child to develop mentalization and symbolic abilities (4). In contrast, children who lack secure attach-

Address for Correspondence: Lily Rothschild-Yakar, PhD, Department of Psychology, University of Haifa, Haifa 31905, Israel [email protected]

35 “Eating Me Up from Inside” ment relationships are at a greater risk for remaining ing and enacting feelings and thoughts, and a means of fixated in primitive modes for representing subjectivity. regulating drives and emotions (11). The domain of mentalization is multifaceted, including Failure in the development of mentalization may lead four dimensions, each constructed of two poles (5): (a) to deficiencies in the development of emotion regulation automatic-implicit mentalization; (b) cognitive vs. affective (1), considered a major deficit in the development and mentalization; (c) metalization based on mental interiors maintenance of EDs (12). It has been shown that patients vs. externally-based mentalization; and (d) mentalizing with EDs exhibit a less developed ability to mentalize self-experience vs. mentalizing the experience of others. experience and to differentiate between perspectives In this study we examine mentalizing self-experience of self and others than non-ED individuals (13, 14). vs. the experience of others, and the general ability of Moreover, patients with anorexia nervosa (AN) present mentalization vs. affective mentalization - based on affec- with elevated level of alexithymia compared to patients tive recognition. with bulimia nervosa (BN), who present with elevated In line with Choi-Kain and Gunderson’s (6) assertion level of alexithymia than controls (15, 16). Regarding that the multifaceted nature of mentalization requires the ToM, whereas some studies have found no deficits in use of a multi-dimensional approach, based on related emotion recognition (17, 18), others have found difficul- concepts, we examined the sub-construct of affective ties in emotion recognition among female adolescents mentalization based on the related concepts of alexi- suffering from AN (19). thymia and Theory of Mind (ToM). Alexithymia, as conceptualized by Bagby et al. (7), refers Affect Regulation and Eating Disorders to the subjective experience of the self. Alexithymia is a Jurist (20, 21) suggests that the sub-construct of mentalizing construct encompassing difficulties in identifying subjec- affectivity is the basis of emotion regulation ability. This tive feelings, describing feelings of the self to others, and ability includes the recognition and expression of emotion, a stimulus-bound, externally oriented cognitive style (7). the process of revisiting emotions of past and present, and Alexithymia relates to the subjective experience of emotion regulation. The sub-construct of mentalizing the self, whereas ToM refers to the understanding of the affectivity has not been operationalized yet; in this study, other. ToM refers to the ability to form representations we have related to two of its subcomponents - recognition about intentional internal mental states such as thoughts, and expression of emotions, with the related construct of feelings, and beliefs (8). There are cognitive aspects and alexithymia. We also have examined the recognition of affective aspects of ToM. The cognitive mechanism emotions of others based on the concept of ToM. includes understanding of the beliefs of others, whereas In contrast to the theory relating the ability to regu- the affective mechanism includes recognition of the late emotions exclusively to parent-infant relationship, emotions of others, and empathizing with them (5). In other researchers suggest that it may be mainly tied to our study we will examine emotion recognition. physiological or cognitive processes (22). The model The relationships between emotional mentalizing, presented in our study combines the two, in postulating ToM, and alexithymia were shown empirically. that whereas the ability to regulate emotions may be For example, healthy individuals with elevated alexi- connected to the development of mentalization, other thymia showed lower levels of empathy and performed innate physiological and cognitive maturation processes less well in tasks of emotion recognition (9). are involved as well. Emotion regulation strategies are the processes trig- Mentalization, Alexithymia, Affective ToM and gered following the attending and evaluating of emotional Eating Disorders cues, involving experiential, behavioral and physiologi- According to the mentalization model, EDs are con- cal systems (23). A developed ability to regulate affect ceptualized as related to primitive modes of experience involves the ability to cope with distress and attenuate (10). People with EDs have failed to develop the ability negative emotion as well as strengthening positive emo- to differentiate between physical and emotional states tions (24). and between one’s own and others’ experiences, and to The process model of emotion regulation suggests ascribe causality to self and interpersonal experiences. that there are two types of emotional regulation strate- These deficiencies may lead to the use of bodily- and gies: reappraisal of emotions and suppression of emo- ED-related symptoms as a concrete means of represent- tions. Cognitive reappraisal involves a cognitive change

36 Lily Rothschild-Yakar et al. regarding the emotion-eliciting situation in a way that aimed to explore the predictive value of mentalization, the changes its emotional impact (for example by using quality of emotion regulation strategies, and depression on inner speech to encourage oneself while facing stress). It the severity of ED symptoms. Our third hypothesis was occurs early and intervenes before the emotional response that comorbidity with depressive symptoms would predict has been fully generated. Expressive suppression is a greater ED symptomatology, whereas greater ability of form of response modulation involving inhibition of general and affective mentalization and adequate emotion ongoing emotion-expressive behavior (25). Reappraisal regulation strategies would serve to attenuate ED symptoms. should successfully reduce the experiential and behavioral components of negative emotion, whereas suppression which comes relatively late in the emotion-generation Method process primarily modifies the behavioral aspect of the Participants emotional response. The ED sample comprised of 25 female inpatients between Gross and John (23) suggest that people using cognitive the ages of 14-24 years who were hospitalized in the ado- reappraisal experience and express more positive and less lescent or adult ED inpatient departments located in a negative emotions and depressive symptoms; by contrast, general hospital in central Israel. The ED sample is part people tending to suppress emotions experience and express of a study assessing mentalization in female adolescents less positive and more negative emotions and depressive and their parents during the years 2014-2016. It represents symptoms. Swart et al. (9) further show that individuals all cases in which both patients and parents agreed to with elevated alexithymia tend to suppress emotion and participate in the study. All patients met the criteria for use cognitive reappraisal to a lesser extent than those with a full-blown DSM-5 diagnosis of ED (27) on admission low alexithymia. Lastly, a meta-analysis about emotion and had never been diagnosed with a bipolar disorder, regulation in several clinical groups has demonstrated schizophrenic spectrum disorder, substance use disorder, significant emotion suppression in EDs, but no difference mental retardation, organic brain syndrome or any physical in the use of cognitive reappraisal vs. controls (26). disorder with the potential to affect appetite or weight (e.g., thyroid disorder or diabetes mellitus). The patients were Aims and Hypotheses of the Study diagnosed with either AN-restricting type (AN-R; n = First, we aimed to examine the relationship of the general 13), AN binge/purge type (AN-B/P; n = 7), or BN (n = 5). ability of mentalization, as well as of the specific domain Whereas the different ED subtypes may differ in several of affective mentalization about self and others, with the personality characteristics (28), there are many similarities strategies of emotion regulation. Based on the suggestions among them in other personality characteristics and in of Fonagy and Target (2), we hypothesized that higher comorbid psychiatric disorders (16, 28). In addition, many mentalizing abilities (high reflecting functioning [RF], patients may oscillate among the various ED diagnoses high affective ToM and low alexithymia) will be positively during the course of their illness (29). Therefore, in this correlated with the cognitive reappraisal strategy, and study, we related to all patients as belonging to one group. negatively correlated with the expressive suppression The control sample comprised of 22 high school and strategy. We also anticipated that the correlational size undergraduate female students who volunteered to par- of affective mentalization measures (affective ToM and ticipate in the study and of other undergraduate female alexithymia) with emotion regulation strategies will be students who participated as part of the requirements for stronger than with the general RF. completing their degree. The control participants were The second aim was to examine the between-group matched to the research group by age, marital status differences in mentalization abilities and emotion regula- (all participants were single), and parents’ educational tion strategies. We anticipated that the ED group would level. Controls were required to have no lifetime history show lower RF, lower ability to identify the emotions of psychiatric illness, physical illness with the potential of others (ToM), and greater alexithymia than the con- to affect appetite or weight, chronic use of medications, trol group. Furthermore, the ED group would use more and any stigmata indicative of an ED. expressive suppression and less cognitive reappraisal To control for intellectual ability, we administered two than the controls. subtests from the Hebrew edition of the Wechsler Adult Depression is found to be more frequent in patients Intelligence Scale (30, 31) as an estimate of Intelligence with EDs vs. controls (16). In line with these findings, we Quotient (IQ): block design and similarities.

37 “Eating Me Up from Inside”

Measures and complex emotion of others. We used the Hebrew Independent Variables: Mentalization version translate by Milo. Participants are presented with The Reflective Function (RF) Scale (3). The RF scale inte- 36 photographs of eyes posing emotions. Participants grates the assessment of multiple facets into a global rating are asked to make a forced decision between 4 words for the quality of mentalizing in the context of childhood naming the emotion (1 correct and 3 distracters). The attachment relationships via narratives derived from test retest reliability and validity of the RME were previ- the Adult Attachment Interview (AAI(32). Interviews ously established (35). are scored on an 11-point rating scale ranging from –1 (negative RF, in which interviews are overly concrete or Independent Variables: Emotion Regulation grossly distorting others’ mental states, e.g., “…but what Strategies makes me feel more rejected is that she breast-fed me, Emotion Regulation Questionnaire (ERQ) (23). The and she didn’t breast-feed my sister”), through ordinary ERQ is a 10-item self-report comprised of two subscales: RF that is common in non-clinical populations (e.g., “I a. cognitive reappraisal strategy (e.g., “I control my emo- think not only my mother held on to me very tightly, tions by changing the way I think about the situation I’m but I wrenched away very hard, and those two actions in”) and b. expressive suppression (e.g., “I control my made it much more extreme than it might have been emotions by not expressing them”). Participants rate items otherwise”), to +9 (exceptional RF, in which interviews on a 7-point scale ranging from 1 (strongly disagree) to show unusually complex, elaborate, or original reasoning 7 (strongly agree). The test retest reliability and validity about mental states of self and others). of the test were previously established (23). (Cronbach In our study we added specific RF subscales regarding alpha in this study for cognitive reappraisal=.86; for the self, mother and father in addition to the general RF expressive suppression=.72). scale, according to specific questions (e.g., self: “Do you think your childhood experiences have an influence on Comorbidity: Depression who you are today?”). These three specific scores are Beck Depression Inventory (BDI) (36) is a 21-item designed according to the guidelines of the general RF self-report assessing the severity of depressive symptoms score. at the time of the evaluation. Participants rate items on Two coders were trained by the first author, who had a 4-point scale ranging from Rarely (1) to Often (4). received training from the developers of the coding (Cronbach alpha for the BDI in this study=.94) manual. One coder scored all the protocols. The two coders, who were blind to the group allocations of the Dependent Variable: ED Symptoms participants, coded a subset of the transcripts (n = 33), The Eating Attitudes Test (EAT-26) (37) is a self-reported yielding excellent inter-rater reliability, intraclass correla- scale assessing concerns and behaviors related to eating. tion coefficient (ICC) for general RF=.97, self RF=.90, Participants rate items on a 6-point scale ranging from mother’s RF=.84, and father’s RF=.88. Never (1) to Always (6) (Cronbach alpha of the EAT-26 TAS-20 – The Toronto Alexithymia Scale (33). The in this study=.96). TAS is a self-report questionnaire assessing difficulties in identifying and communicating feelings of self and a con- Procedure crete/externally-oriented style of thinking. The 20 items All participants, and their parents or other legal guardians are divided into three subscales: difficulty in identifying for minors under age 18, gave their written informed feelings (DIF, 7 items), difficulty in describing feelings consent to participate in the study after receiving an expla- (DDF, 5 items) and externally oriented thinking (EOT, nation of the study’s goals and methodology. The study 8 items). The reliability and factorial validity of the TAS was approved by the Internal Review Boards (Helsinki have been established previously (33). Participants rate Boards) of the Sheba Medical Center. items on a 6-point scale ranging from strongly disagree Research group. Experienced child and adolescent (1) to agree (6). Cronbach alpha in this study for the full and adult psychiatrists (DS, AHL, EG) established the scale =.85; for DIF=.89, for DDF=.84, and for EOT=.53. diagnosis of ED within 14 days of admission using an Complex Emotion Recognition and Theory of Mind adapted version of the Structured Clinical Interview for – Reading the Mind in the Eyes Task (RME) (34). The Axis I DSM-IV Disorders – Version 2.0 (SCID-I I/P, task was designed to assess the ability to recognize basic Version 2.0) (38). The degree of inter-rater reliability

38 Lily Rothschild-Yakar et al. values (according to the correlation coefficient procedure) father’s education in the statistical analyses. As expected, among the three psychiatrists for the diagnosis of an ED patients with EDs reported significantly more ED symptoms was r=.92. All ED diagnoses were confirmed in clinical and depression compared to the non-ED control group. team meetings of the respective inpatient departments. A comparison of the three ED subtypes revealed no Inclusion criteria were BMI within normal limits (18.5-25 significant differences for age, verbal IQ, non-verbal IQ, kg/m2) (39), no cessation of menstruation since menarche, mother’s education level, ED symptoms and depression and no EDs and other reported psychopathology. Four (results not shown). As expected, a significant between-group control participants were excluded from the study. difference was found for BMI, with the BN presenting with The study measures were administered individually by a significantly higher BMI than both the AN-R and AN-B/P two trained master’s level clinical psychology interns who groups (AN-R: M=18.91±3.47; AN-B/P: M=20.48±3.82; BN: were blind to patient subgroup affiliation and to the self- M=24.29±3.51; F(2,23)=3.78, p<.05). A significant between- report results. Patients were examined only after the ED group difference was also found for father’s education symptoms and overall medical condition were stabilized, as level (AN-R: M=13.5±2.01; AN-B/P: M=14.82±3.06; BN: determined by physical examinations and relevant labora- M=10.8±1.3; F(2,23)=4.66, p<.05). We therefore controlled tory tests, to reduce the influence of the ED symptomatology for father’s education in the statistical analysis. and malnutrition on the findings of the study. We therefore compared the study measures of the three Control group. To screen out participants with ED ED subtypes using multivariate analyses of covariance symptoms or other psychopathology, the control partici- (MANCOVAs) with father’s education level as a covari- pants reported their weight and height, and whether they ate. No significant difference emerged for mentalization had ever had an ED or any other medical or emotional abilities with the dependent variables: general RF, affec- morbidity. In addition, they filled out the SCOFF ques- tive ToM (RME) and alexithymia (TAS) (Wilk’s Lambda tionnaire for the assessment of eating disorders (40). F(6,36)=0.96, n.s) as well as with the variables: self-RF, The SCOFF is a 5-yes/no question tool, developed to mother-RF, father-RF, RME and TAS (Wilk’s Lambda screen for EDs in the general population. A total score F(10,32)=0.95, n.s). No significant between-group difference of two and above has been found to be 100% sensitive emerged also for emotion regulation strategies (Wilk’s and 87.5% specific for the presence of any ED. Lambda F(4,42) = 0.54, n.s.). These results and the lack of differences in the demographic and symptomatic measures between the three ED subgroups allowed us Results to relate to all patients as belonging to one group. A comparison of the demographic characteristics of two We further examined the correlation between the study groups using a t test is presented in Table 1. The data different measures of general mentalization (general RF, revealed no significant differences for age, mother’s educa- self-RF, mother-RF and father-RF), and the specific sub- tion level, BMI, and estimated measures of IQ. Differences components of affective mentalization - the affective ToM were found in father’s education level, so we controlled for (RME) and alexithymia (TAS) using Pearson correlation coefficients. Elevated general RF score was significantly Table 1. Demographic and Clinical Variables in Patients with correlated with elevated accurate emotion recognition EDs and Controls (RME r=.25, p<.05), and attenuated level of alexithymia Patients Non-ED (TAS r=-.27, p<.05). Self-RF was significantly correlated with EDs controls with TAS (r=-.32, p<.05); father RF was significantly (n = 25) (n = 22) Cohen’s Variables M ± SD M ± SD t (47) d correlated with accurate RME (r=.25, p<.05). Age 17.22±2.77 16.52±1.21 1.19 0.35 Similarity- IQ 10.67±2.43 11.68±2.84 1.2 0.35 Mentalization and Emotions Regulation Blok design - IQ 10.56±3.12 11.81±2.32 1.4 0.41 Strategies Mother’s education 14.54±2.75 15.09±2.02 0.8 0.23 Father’s education 13.58±2.72 16.05±2.24 3.45** 1.01 We examined the first research hypothesis anticipating Body mass index 20.35±3.97 20.71±1.92 0.42 0.12 that there will be correlations between the various aspects ED symptoms (EAT-26) 50.81±14.73 9.64±8.81 12.11*** 3.53 of mentalization and emotion regulation strategies using Depressive symptoms (BDI) 37.48±13.07 5.41±8.63 10.29*** 3.00 Pearson correlation coefficients (see Table 2). The hypoth- **p<0.01 ***p<0.001 BDI= Beck’s Depression Inventory esis was supported for expressive suppression but only partially for cognitive reappraisal. Specifically, greater

39 “Eating Me Up from Inside”

2 F(3,40)=7.26, p< 001, η =0.35), self-RF, mother-RF, father-RF, Table 2. Pearson Correlations between Mentalization 2 Measures (General RF, Mother RF, Father RF, Self RF, RME, RME and TAS (Wilk’s Lambda F(5,38)=5.71, p<.001, η =0.43), TAS) and Emotion Regulation Strategies (n=47) and expressive suppression and cognitive reappraisal (Wilk’s 2 Mentalization Lambda F(2,44)=6.07, p<.01, η =0.22) were significant. The Expressive Cognitive univariate ANCOVAs presented in Table 3 indicate four Emotion Regulation Strategies suppression reappraisal significant between-group effects: self-RF, alexithymia, General Reflective Function (RF) -0.23† -0.07 expressive suppression and cognitive reappraisal, with small Mother’s Reflective Function 0.01 -0.07 to medium effect sizes. Specifically, female adolescents Father’s Reflective Function -0.06 -0.23† with EDs presented lower self-RF, elevated alexithymia, Self-Reflective Function -0.37** -0.01 elevated use of expressive suppression and attenuated use b ToM- Complex Emotion Recognition (RME) -0.23† 0.0 of cognitive reappraisal compared with the controls. Alexithymia (TAS)a 0.48*** -0.49***

†p<0.06 **p<.01 ***p<.001 Mentalization, Emotion Regulation Strategies RF=Reflective Function; RME=Reading the Mind in the Eyes; TAS= The Toronto Alexithymia Scale and Depression as Predictors of ED Symptoms an=46 bn=46 First, we calculated Pearson correlations of the different mentalization abilities (RF, ToM and alexithymia), and reported alexithymia (TAS) was significantly correlated the emotion regulation strategies, with ED symptoms with elevated expressive suppression and with reduced (see Table 4) and depression (BDI). cognitive reappraisal. Elevated self-RF was significantly In the general sample, significant correlations were found correlated, and general RF and RME were marginally cor- between elevated ED symptoms and elevated alexithymia, related with a lower tendency to suppress emotions. These low level of self- RF, attenuated use of cognitive reappraisal data show that female adolescents suppressing their emo- and elevated emotion suppression. Low general RF was tions show an attenuated level of general RF and specifically marginally correlated with elevated ED symptomatology. low level of self-RF, low level of RME and elevated level of In patients with EDs, reduced alexithymia and elevated alexithymia. Interestingly, cognitive reappraisal was only ability of emotion recognition in others (RME) significantly marginally correlated with elevated father-RF. Mother-RF correlated with attenuated ED symptoms. Interestingly, was not related to the emotion regulation strategies. in the control group elevated RME was correlated with The second part of the first hypothesis suggesting that the greater reported ED symptomatology. correlations between measures of affective mentalization Significant correlations were also revealed between (RME, TAS) and emotion regulation strategies will be stron- BDI and self-RF (r=-0.27, p<.05), TAS (r=.73, p<.001), ger than the correlations with the general RF was examined using Fisher r-to-z transformation. The hypothesis was Table 3. Two-Way Analyses of Variance of the Study Variables for the Two Groups partially confirmed. Specifically, the correlation between Patients Non- alexithymia and cognitive reappraisal was significantly with EDs patients F stronger than the correlation between general RF and cogni- (n = 25) (n = 22) (1, 46) Eta² tive reappraisal (Fisher’s Z=2.31, p<0.05). Nonetheless, no Variable M SD M SD significant differences were detected between alexithymia General Reflective Function 3.88 1.88 4.91 1.34 1.02 .02 (RF) X expressive suppression and general RF X expressive Mother’s Reflective Function 3.85 1.49 4.45 1.50 0.65 .01 suppression (Fisher’s Z=0.96, n.s). No significant differ- Father’s Reflective function 3.85 1.85 4.32 1.52 0.01 .00 ences emerged regarding the correlations between RME Self-Reflective Function 3.92 1.7 5.50 1.10 6.05* .12 and expressive suppression and between general RF and ToM- Complex Emotion 24.04b 3.59 24.48 3.44 0.05 .001 expressive suppression (Fisher’s Z=0.61, n.s). Recognition (RME) Alexithymia (TAS) 65.53 13.21 47.14a 9.42 23.5*** .35 Between-Group Differences in Mentalization and Cognitive Reappraisal 3.76 1.71 4.84 0.8 9.28** .17 Emotion Regulation Strategies Expressive Suppression 4.01 1.43 2.86 1.01 6.66* .13 To examine the second hypothesis, we conducted three *p<0.05 **p<.01 ***p<0.001 RME=Reading the Mind in the Eyes; TAS= The Toronto Alexithymia Scale between-group MANCOVAs with father’s education as a RF=Reflective Function; RME=Reading the Mind in the Eyes; TAS= The covariate. The MANCOVAs for all three groups of depen- Toronto Alexithymia Scale a b dent variables: General RF, RME and TAS (Wilk’s Lambda n=21; n=24

40 Lily Rothschild-Yakar et al.

Table 4. Pearson Correlations of Mentalization and Emotion education was included in the first step. In the second Regulation Strategies with ED Symptoms step, the mentalization measures were entered in a step- Total Patient Control wise regression procedure; in the third step the emotion Group Variables sample with EDs group regulation strategies were added, and in the fourth step General Reflective Function (RF) -.33† -.31 .01 depression was added in a stepwise regression procedure. Mother’s Reflective Function -.15 .09 -.25 The regression analysis was found to be highly signifi- Father’s Reflective Function -.17 -.29 -.01 cant, F(7,38)=18.83, p<.001, explaining 74.8% of the variance Self- Reflective Function -.45*** -.12 -.05 in the EAT-26. Specifically, in the first step father’s educa- ToM-Accurate emotion recognition -.16 -.45*b .37* tion contributed 18.6% of the variance. The second model (RME) comprising the mentalization measures block in addition Alexithymia (TAS) .70*** .58** -.08a Cognitive Reappraisal -.50*** -.27 -.25 to father’s education was found as significant F(4,41)=13.24, Expressive Suppression .44*** .19 -.23 p<0.001, significantly explaining an additional 39.6% of the variance. The sole significant variable in this block †p<.06; *p<0.05; **p<0.01; ***p<0.001 RME=Reading the Mind in the Eyes; TAS= The Toronto Alexithymia Scale was the alexithymia measure (TAS) showing that elevated an=21; bn=24 TAS contributed to the severity of ED symptomatology.

The third model was also significant F(6, 39) = 9.33, p<0.001, cognitive reappraisal (r=-.48, p<.001), expressive sup- but the addition of emotion regulation strategies was not pression (r=.53, p<.001) and EAT-26 (r= .86, p<.001). significant; father’s education and alexithymia retained Specifically, elevated self-RF, low alexithymia, greater use their significance in the prediction of ED symptoms, and of cognitive reappraisal and reduced emotion suppression cognitive reappraisal was marginally significant. In the were found as related to attenuate depressive symptoms. fourth model, the inclusion of depression measure was a significant predictor, raising the explained variance to Prediction of the Severity of ED Symptoms 74.8%. In this mode depression was the sole significant A linear regression analysis was conducted to examine the variable, predicting greater severity of ED symptomatol- third research hypothesis regarding the potential of the ogy, and elevated general RF was marginally significant different mentalization abilities (RF, ToM and alexithymia) predictor of attenuated ED symptoms. and emotion regulation strategies to predict the severity of ED symptomatology (EAT-26) (see Table 5). Father’s Discussion Table 5. Prediction of ED Symptoms by Mentalization, We examined the relationship between general ability Emotion Regulation Strategies and Depression of mentalization, affective mentalization regarding self 2 Step Predictors β ∆R Fchange and others, strategies of emotion regulation, and ED 1 Father’s education -0.43** 0.19 9.35** symptoms in female adolescent inpatients with EDs vs. 2 Father’s education -0.27** 0.4 12.02*** controls. The underlying assumption was that patients General RF -0.07 ToM-Accurate emotion recognition (RME) -0.01 with EDs would present with a deficiency in mentalization Alexithymia (TAS) 0.63*** and in adopting mature emotion regulation strategies. 3 Father’s education -0.29** 0.03 1.22 According to Fonagy and colleagues (1), deficiencies General RF -0.04 ToM-Accurate emotion recognition (RME) -0.05 in the development of mentalization impair the abil- Alexithymia (TAS) 0.52** ity to develop adequate emotion regulation strategies. Cognitive Reappraisal -0.22† According to this assumption, EDs may develop as a Expressive Suppression 0.06- means to build a sense of cohesiveness and self-regulation 4 Father’s education -0.13 0.18 30.29*** General RF -0.16† (41). In addition, we designed specific general RF scores ToM-Accurate emotion recognition (RME) 0.02 for the self, mother and father to explore the specific Alexithymia (TAS) 0.07 dimension of mentalization regarding self vs. others, Cognitive Reappraisal -0.12 Expressive Suppression -0.14 rather than only examining the general ability of RF. BDI 0.70*** Preliminary analyses showed significant correlations †p=.09, p *<0.05, ** p <0.01, *** p <0.001 between different mentalization abilities. Thus, higher RF=Reflective Function; RME=Reading the Mind in the Eyes; BDI= levels of general RF and RF regarding the self were cor- Beck’s depression Inventory related with elevated accuracy in identifying the emotions

41 “Eating Me Up from Inside” of others (RME), as well as with lower levels of alexithymia reappraisal than controls, with small to medium effect (TAS). These data are in accordance with other studies sizes. There were no between-group differences in RME, showing an overlap between the different constructs and and in general and parents’ RF. mentalization abilities (9). The lack of between-group difference in RME in our Interestingly, no significant association was found study is in contrast to studies showing differences in between the subjective ability to recognize and report RME between patients with AN and controls (42) but in affects regarding the self (TAS) and the objective accuracy accordance with other studies showing no RME deficits in recognizing emotions in others (RME). This find- in patients with BN (17). Furthermore, our data show ing is in contrast to neuroimaging studies showing that that the groups differ in mentalizing regarding the self thinking on emotions of others evokes the same areas of (self-RF, alexithymia) but not in mentalizing others’ recognition and reflection on the emotions of the self (5). experiences (parents-RF, RME). This self/other gap is in Nonetheless, these data may support the description that accordance with claims of Skårderud and Fonagy (41) that the ability to mentalize may vary across different con- subjects with EDs do not develop a coherent sense of self texts - people may be sensitive to cues regarding affective from within; rather, they invest in getting a sense of self states of others while having difficulties in being aware based on the reactions of others. This gap corresponds of their inner affective states. An alternative explanation also with self-psychology conceptualization and data of the self/other discrepancy may rely on the influence showing that the development and maintenance of an of the assessment procedure. Thus, whereas alexithymia ED may be associated with attunement to the needs of is measured in this study by the subjective self-report of the other, and not to one’s own perspective (43). the TAS, emotion recognition in others is measured by The results regarding the between-group differences an objective measure, the RME. in emotion regulation strategies are in accordance with previous data showing that patients with EDs tend to Mentalization and Emotion Regulation Strategies use inadequate emotion regulation strategies, in particu- The correlations found in our study between mentaliza- lar expressive suppression (26). As has been previously tion measures and emotion regulation strategies partially described, cognitive reappraisal entails changing the way support our first hypothesis, in showing that participants a situation is construed, aiming to decrease the experi- using more expressive suppression present marginally ential and behavioral components of negative emotion. significant lower levels of general RF, significantly lower The negative correlation between cognitive reappraisal levels of self-RF and higher levels of alexithymia (TAS). and BDI, found in our study, is in line with this claim However, cognitive reappraisal has shown a significant and with previous studies (23). Our findings suggest that negative correlation only with alexithymia. These cor- attenuated use of this adaptive regulation strategy and relations between alexithymia and the different emotion greater use of emotion suppression are correlated with regulation strategies replicate findings of Swart et al. (9) elevated depression and may lead to highly inefficient study in healthy controls. consequences, such as bingeing or other disordered eat- The finding that mentalization measures are correlated ing, to alleviate their negative feelings. to a greater extent with expressive suppression than with cognitive reappraisal suggests that mentalization Mentalization, Emotion Regulation Strategies processes in our sample have an influence on a later and Depression as Predictors of ED Symptoms stage of emotion processing – the response phase, rather The third hypothesis was that elevated mentalization and than on earlier stages, before the emotional response has adequate emotion regulation predict the attenuation of been fully generated. the severity of ED symptoms. By contrast, comorbidity with depression would predict elevated ED symptoms. Between-Groups Differences in Mentalization As expected, the preliminary data showed significant and Emotion Regulation Strategies correlations between ED symptoms and general RF, self- The second hypothesis anticipating between-group differ- RF, alexithymia, and both emotion regulation strategies. ences in mentalization measures and emotion regulation These findings and further data showing that in patients was partially confirmed. Participants with EDs showed with EDs, alexithymia and elevated level of accurate lower levels of self-RF and higher levels of alexithymia, emotion recognition in others are related to attenuated and used more expressive suppression and less cognitive ED symptoms, lend support to this hypothesis.

42 Lily Rothschild-Yakar et al.

However, a more detailed examination of our data (e.g., express feelings to them). This approach may replace using linear regression analyses showed that for the men- the tendency of using the body and ED-related symptoms talization block, alexithymia was the sole variable predict- as a concrete means of representing and enacting feelings ing the severity of ED symptoms, i.e., elevated alexithymia and thoughts (e.g., “when my parents quarrel I work out predicted more severe ED symptoms. Reduced use of until my body aches”). Additionally, treatment should cognitive reappraisal also predicted greater ED symptom- help patients with EDs to regulate their emotions more atology. Most importantly, upon introducing depression adaptively with the use of the cognitive reappraisal pro- into the regression model, there was no longer a distinct cess instead of suppressing their emotions; in that case, contribution of alexithymia regarding the prediction of reappraisal of the emotion-eliciting negative situation the severity of ED symptoms. These data may be related (e.g., “My parents quarrel now but I remind myself that to the common variance between alexithymia and depres- they usually work it out and solve their problems”) may sion found in our study and elsewhere (44). A common change its negative impact. variance was also found between depression and cognitive Future research should be longitudinal and prospec- reappraisal; thus, when depression was included in the tive, including a larger number of ambulatory patients, regression analysis, this emotion regulation strategy no and examining whether there would be an association longer predicted the severity of ED symptomatology. between improvement in ED symptoms and improve- Our data further indicated that elevated general RF, ment in mentalization and emotion regulation over the which includes both affective and cognitive aspects, was course of the illness. a marginal significant contributor to the prediction of attenuated ED symptoms, independent of the predictive Contribution of the authors power of depression. The first, second and last authors substantially contributed to all phases including conception and design, analysis and interpretation of data, writing and final approval. The third and fourth authors contributed substantially to Limitations, Conclusions and Recommendations the administration, analysis, interpretation of data and critical revision. for Future Research The authors declare that they have no financial support or relationship that First, our research design is cross-sectional, allowing may pose a conflict of interest. only for inferences about associations but not causation. Second, our sample was relatively small, precluding the References separate evaluation of each ED subtype. Lastly, as our 1. Fonagy P, Gergely G, Jurist E, Target M. Affect regulation, mentalization, sample included only inpatients, our findings cannot be and the development of the self. New York, N.Y.: Other Press, 2002. generalized to less severe EDs 2. Fonagy P, Target M. An interpersonal view of the infant. In Hurry A, editor. and developmental therapy. London, U.K.: Karnac Despite the aforementioned limitations, our data sup- Books, 1998: pp. 3-31. port the suggestion to use a multidimensional approach 3. Fonagy P, Target M, Steele H, Steele M. Reflective-functioning manual, based on different measures for the assessment of met- version 5.0, for application to adult attachment interviews. London, allization (6). In our study, we assessed mentalizing U.K.: University College London, 1998. 4. Taylor GJ. Affects, trauma, and mechanisms of symptom formation: A regarding self vs. other and subcomponents of affective tribute to John C. Nemiah, MD (1918-2009). Psychother Psychosom mentalization, and found a profile that may specifically 2010;79:339-349. characterize patients with EDs. 5. Fonagy P, Luyten P. A developmental, mentalization-based approach to the understanding and treatment of borderline personality disorder. The finding that high mentalization ability bears an Dev Psychopathol 2009;21:1355-1381. attenuating effect on ED symptoms, whereas depression 6. Choi-Kain LW, Gunderson JG. Mentalization: Ontogeny, assessment, predicts a high level of these symptoms, may have clinical and application in the treatment of borderline personality disorder. Am J Psychiatry 2008;165:1127-1135. implications. Treatment of EDs should focus not only 7. Bagby RM, Quilty LC, Taylor GJ, et al. Are there subtypes of alexithymia? on the behavioral aspects of the disorder, but should Person Individ Differences 2009;47:413-418. also assist patients in improving their ability to think 8. Heavey L, Phillips W, Baron-Cohen S, Rutter M. The Awkward Moments reflectively about themselves and others, and to identify Test: A naturalistic measure of social understanding in . J Autism Dev Disord 2000;30:225-236. and express their emotions (41). Gaining the ability 9. Swart M, Kortekaas R, Aleman A. Dealing with feelings: Characterization to identify mental tension and express it verbally and of trait alexithymia on emotion regulation strategies and cognitive- symbolically and further reflecting on causality of these emotional processing. PLoS One 2009;4:e5751. 10. Skårderud F. Eating one’s words, Part I:‘concretised metaphors’ and mental states (e.g., “I feel helpless and angry when my reflective function in anorexia nervosa – An interview study. Europ parents quarrel”) may lead to better coping with stressors Eat Disord Rev 2007;15:163-174.

43 “Eating Me Up from Inside”

11. Bleiberg E. Treating personality disorders in children and adolescents: 29. Fichter MM, Quadflieg N, Hedlund S. Long‐term course of binge eating A relational approach. New York, N.Y.: Guilford Press, 2004. disorder and bulimia nervosa: Relevance for nosology and diagnostic 12. Gilboa‐Schechtman E, Avnon L, Zubery E, Jeczmien P. Emotional criteria. Int J Eat Disord 2008;41:577-586. processing in eating disorders: Specific impairment or general distress 30. Wechsler D, Kaplan E, Fein D, et al. Wechsler Intelligence Scale for related deficiency? Depress Anxiety 2006;23:331-339. Children Fourth Edition–Integrated: Technical and interpretative manual. 13. Rothschild-Yakar L, Levy-Shiff R, Fridman-Balaban R, Gur E, Stein San Antonio, Texas: Psychological Corporation, 2004. D. Mentalization and relationships with parents as predictors of eating 31. Wechsler D. WAIS-III: Administration and scoring manual: Wechsler disordered behavior. J Nerv Ment Dis 2010;198:501-507. adult intelligence scale. San Antonio, Texas: Psychological Corporation, 14. Ward A, Ramsay R, Turnbull S, et al. Attachment in anorexia nervosa: 1997. A transgenerational perspective. Br J Med Psychol 2001;74:497-505. 32. George C, Kaplan N, Main M. The Adult Attachment Interview. 15. Corcos M, Guilbaud O, Speranza M, et al. Alexithymia and depression Unpublished manuscript, University of California, Berkeley (third in eating disorders. Psychiatry Res 2000;93:263-266. edition), 1996. 16. Speranza M, Corcos M, Loas G, et al. Depressive personality dimensions 33. Bagby RM, Parker JD, Taylor GJ. The twenty-item Toronto Alexithymia and alexithymia in eating disorders. Psychiatry Res 2005;135:153-163. Scale – I. Item selection and cross-validation of the factor structure. J Psychosom Res 1994;38:23-32. 17. Dejong H, Van den Eynde F, Broadbent H, et al. in bulimia nervosa: A systematic review. Europ Psychiatry 2013;28:1-6. 34. Baron‐Cohen S, Wheelwright S, Hill J, Raste Y, Plumb I. The “Reading the Mind in the Eyes” test revised version: A study with normal adults, 18. Rothschild‐Yakar L, Eviatar Z, Shamia A, Gur E. Social cognition in and adults with Asperger syndrome or high‐functioning autism. J Child eating disorders: Encoding and representational processes in binging Psychol Psychiatry 2001;42:241-251. and purging patients. Europ Eat Disord Rev 2011;19:75-84. 35. Vellante M, Baron-Cohen S, Melis M, et al. The “Reading the Mind 19. Harrison A, Sullivan S, Tchanturia K, Treasure J. Emotion recognition in the Eyes” test: Systematic review of psychometric properties and a and regulation in anorexia nervosa. Clin Psychol Psychotherapy validation study in Italy. Cog Neuropsychiatry 2013;18:326-354. 2009;16:348-356. 36. Beck AT. A systematic investigation of depression. Compr Psychiatry 20. Jurist EL. Minds and yours: New directions for mentalization theory. 1961;2:163-170. In Jurist EL, Slade A, Bergner S, editors. Mind to mind: Infant research, 37. Garner DM, Olmsted MP, Bohr Y, Garfinkel PE. The eating attitudes test: neuroscience, and psychoanalysis. New York, N.Y.: Other Press, 2008: Psychometric features and clinical correlates. Psychol Med 1982;12:871-878. pp. 88-114. 38. First MB, Spitzer RL, Gibbon M, Williams, JB. Structured clinical 21. Jurist EL. Mentalizing minds. Psychoanal Inquiry 2010;30(4):289-300. interview for DSM-IV axis I disorders- subject edition (SCID-I/P, 22. Eisenberg N, Spinrad TL. Emotion‐related regulation: Sharpening the version 2.0). New York, N.Y.: Biometric Research Institute, New York definition. Child Dev 2004;75:334-339. State Psychiatric Institute, 1995. 23. Gross JJ, John OP. Individual differences in two emotion regulation 39. Bray GA. Pathophysiology of obesity. Amer J Clin nutrition processes: Implications for affect, relationships, and well-being. J Pers 1992;55:488S-494S. Soc Psychol 2003;85:348-362. 40. Morgan JF, Reid F, Lacey JH. The SCOFF questionnaire: Assessment 24. Schore AN. Effects of a secure attachment relationship on right brain of a new screening tool for eating disorders. BMJ 1999;319:1467-1468. development, affect regulation, and infant mental health. Infant Ment 41. Skårderud F, Fonagy P. Eating disorders. In: Bateman AW, Fonagy P, Health J 2001;22:7-66. editors. Handbook of mentalizing in mental health practice. Washington 25. Gross JJ. Antecedent- and response-focused emotion regulation: DC: American Psychiatric Association, 2012: pp. 347-383. Divergent consequences for experience, expression, and physiology. 42. Harrison A, Sullivan S, Tchanturia K, Treasure J. Emotional functioning J Pers Soc Psychol 1998;74:224-237. in eating disorders: Attentional bias, emotion recognition and emotion 26. Aldao A, Nolen-Hoeksema S, Schweizer S. Emotion-regulation strategies regulation. Psychol Med 2010;40:1887-1897. across psychopathology: A meta-analytic review. Clin Psychol Rev 43. Bachar E, Gur E, Canetti L, Berry E, Stein D. Selflessness and perfectionism 2010;30:217-237. as predictors of pathological eating attitudes and disorders: A longitudinal 27. Diagnostic and Statistical Manual of Mental Disorders, 5th ed. study. Europ Eat Disord Rev 2010;18:496-506. Washington DC: Psychiatric Association, 2013. 44. Honkalampi K, Hintikka J, Tanskanen A, Lehtonen J, Viinamäki H. 28. Godt K. Personality disorders in 545 patients with eating disorders. Depression is strongly associated with alexithymia in the general Europ Eat Disord Rev 2008;16:94-99. population. J Psychosom Res 2000;48:99-104.

44