Fears of Compassion in a Depressed Population Implication for Psychotherapy

Background: While psychological therapies for depression have advanced in the last 20 years, still many people respond only partially and remain vulnerable to relapse. Insight into the limitations of our psychological therapies might be obtained from recent research that has revealed, in nonclinical populations, that some people can be fearful of positive emotions especially affiliative and compassion-focused ones. Aims: This study explores the fears of compassion in a clinical population and their associations with selfcriticism, self-compassion and depression, anxiety and stress. Method: 53 depressed patients completed a series of self-report scales. Results: Fears of compassion, particularly for oneself and from others, were strongly linked to self-criticism, depression, anxiety and stress, and negatively associated with self-compassion and self-reassurance. Conclusions: Since compassion and the affiliative emotions associated with compassion play a fundamental role in emotion regulation, individuals who are blocked or fearful of accessing these emotions are likely to be struggle with emotional regulation and the psychotherapeutic process. Research on the fears of compassion and affiliative emotions suggests these are important therapeutic targets.


Introduction
A recent meta-analysis has suggested that there are few differences of efficacy between different therapies for depression [1]. Although combinations of drugs and psychotherapy perform slightly better than individual therapies alone, many depressed patients can still fare poorly, with relapse the longer people are followed up [2]. Indeed, it is recognised that our psychological treatments for depression are far from adequate with perhaps only 50% making full recoveries, with risk of relapse remaining high and many patients continuing to have residual symptoms. Problems in social functioning can persist well after symptoms have reduced [3].
One of the reasons that therapies may not be as good as therapists want them to be is because there has been no agreed process that therapy for depression should focus on. Psychodynamic therapies, for example, suggest that therapists address unconscious processes especially that of anger [4]. Behavioural therapies argue that we need to increase behavioural repertoires and cognitive therapists focus on challenging dysfunctional beliefs and thoughts [5]. However, recently attention has focused on the role of cultivating compassion for self and others. This is associated with generating positive emotions, particularly those associated with affiliation, connectedness and attachment. Indeed, there is now increasing evidence that helping people develop compassionate and affiliative feelings for others, and oneself, significantly contributes to mental health, well-being and social harmony [6][7][8][9][10] and has a range of neurophysiological effects [11,12]. Compassion focused therapy (CFT) pays particular attention to the ability to generate affiliative motives, behaviours and emotions [13,14]. Focusing on the experience and development of compassion has been found to reduce depression, anxiety and self-criticism in people presenting to a community mental health team [15], in people with long term mental health problems [16], and people in a high security psychiatric setting [17]. CFT has been shown to be helpful for people with psychosis [18,19] and can significantly reduce paranoid ideation [20] and emotional difficulties for people with personality disorder difficulties [20]. Ashworth et al. [21] found CFT to be a valuable addition to standard interventions in helping people with acquired brain injury. Kuyken et al. [22] found that in a mindfulness trial, self-compassion was a significant mediator between mindfulness change and recovery from depression.
This suggests that research needs to explore the nature of compassion, how it is developed and cultivated, and its impact as a therapeutic intervention. When used as a psychotherapy, the cultivation of compassion has different components according to whom it is directed. These are the capacity to experience compassion for others, compassion from others and self-compassion.

Compassion for others
Seeking enlightenment, in order to benefit others, is the focus for a particular kind of self-identity within the Mahayana Buddhist traditions called Bodhicitta, [23,24]. In these traditions cultivating compassion for others is regarded as central to the development of enlightened well-being [23,24]. Indeed, research is confirming this with an increasing number of studies showing that compassionate meditations focused on others (as well as on the self) influence wellbeing, coping, social connectedness [8,10,25,26] and affects neurophysiological systems, especially the frontal cortex [11,12]. Directing people to feel compassionate attitudes towards another's suffering, activates the mesolimbic neural system [27]. Developing a self-compassionate identity (wanting to be helpful to others), in contrast to an ego focused identity (wanting to be recognised and avoid being shamed), is associated with better and more supportive social relationships, commitment and well-being [28]. Depressive symptoms can also reduce with positive psychology interventions such as random acts of kindness [29]. There is also increasing evidence that taking an interest in, and helping others, is associated with positive emotion even in young children [30].

Compassion from others
A second focus for compassion is our openness and responsiveness to the compassion and care we receive from others. It is now known that being the recipient of compassionate and affectionate caring throughout life, and especially in childhood, influences us in many ways. Early experiences of caring affect genetic expression and reduces to depression [31,32]. Early experiences of compassionate caring also have a range of effects on physiological systems that significantly affect the maturation of emotion-regulation systems [33], our sense of ourselves and personal identity, and our values and abilities to be compassionate to self and others [34]. In fact, being receptive to care and receiving social support, validation, kindness and compassion from others, plays a major role in well-being throughout life [35] and buffers against mental health problems [35,36] including depression. Social affiliation can improve recovery and prevent relapse [37]. Compassionate and affiliative support has direct effects on stresslinked amygdala processing [38] and frontal cortical function [39,40]. So the ability to share positive emotions in relationships, including kindness and caring, is important for well-being and relationship quality [41].

Compassion for self
It has long been known that self-dislike and self-criticism are significant vulnerability factors for depression [42] whereas developing affiliative and compassionate feelings and attitudes towards oneself are associated with well-being and coping [9,13,43]. Based on her own definition of self-compassion and measure, [43][44][45][46] Neff found that self-compassion can be distinguished from self-esteem and predicts some aspects of well-being better than self-esteem. Neff and Beretvas [47] found that self-compassionate individuals had better functioning relationships and satisfaction compared to low self-compassionate people. Self-compassion aids coping with academic failure [48,49] and compassionate letter writing to oneself improves coping with life events and reduces depression [50]. Kuyken et al. [22] found that selfcompassion was a significant mediator in the effectiveness of mindfulness-based, cognitive therapy for depression.

Fears, blocks and resistances to compassion
There is good evidence then to suggest that focusing on the cultivation of compassion, such that individuals have more affiliative relationships with themselves and others, could well be an important process in the development of psychotherapy for depression. Indeed, compassion may well have an important role to play in rebalancing attachment-based difficulties [14]. However, as CFT developed, it became clear that there are important blocks, fears, and resistances to the concept of compassion and the experience of affiliative emotion [51]. Gilbert et al. developed a measure of 'fears of compassion' and found them to be highly correlated to psychopathology symptoms. In a follow-up study, Gilbert et al. [52] found fears of compassion to be linked to fears of happiness in general, and problems with emotional processing (alexithymia) and mindfulness. This studies of the fears of compassion to date however have been with non-clinical populations. The study looks at these different fears of compassion in a clinical population.
There are in fact many reasons why people can become fearful of compassion and affiliative emotion and block it, there by increasing their vulnerability to mood disorders and reducing recovery. One reason is they see compassion as a weakness or indulgence and require clarity on the definition and process. Another reason is children who have experienced abuse and neglect may have these emotional memories reactivated by cues of kindness, because they stimulate the attachment system and hence trigger whatever memory is coded there. Indeed, people from traumatic backgrounds can have what has been called "attachment phobia", and a fear affilative emotion [53].
Liotti [54] outlined how abusive and/or neglectful parenting creates approach-avoidance conflicts. For example, for most children, the source of distress resolution is to return to the parent for comfort, support and understanding. However, unpredictable and abusive parents, who are poor at mentalizing and empathy, can also be the source of threat to the child. So the natural (innate) system of seeking comfort is blocked because although the parent and be a source of safeness he/she can be the very threat the child needs to escape from. Liotti [55] describes how those children who are threatened by their parents, and therefore cannot use them as a soothing object, enter states of "threat without resolution." In these conditions the only resolution may be withdrawal and avoidance and trying to close down. These kinds of fears can be understood using classical conditioning of emotional memories [56]. Hence, not only do these individuals have few positive memories of affection and joyfulness they may also have few memories of comfort and soothing when under stress, and instead have many aversive memories associated with wanting or seeking closeness, kindness and compassion.
In contrast, some authors suggest that people can be fearful of compassion and affiliative feelings if they are also fearful of their own damaging potential for rage [57] -a scenario beautifully enacted in the film "Good Will Hunting". A typical scenario for a patient would be "If you really knew me, and what went on in my mind you would not think I deserve compassion." So a sense of shame, and believing that there are "bad things (feelings, thoughts and fantasies) inside of one" which cannot be revealed, maybe a major block to being open to affiliation. So the ability to be open to affiliation may first require individuals to acknowledge and work on their rage; compassion and feelings of affiliation may be quite difficult in the context of a lot of unprocessed rage [4]. Here the therapist teaches 'compassion for and with rage'.
Another mechanism involved in fear of compassion and affiliative feelings is unprocessed grief. Gilbert and Irons [58] suggested that some people are in a state of frozen grief. If therapeutic compassion and kindness begin to activate the attachment and affiliative systems, the feelings of loneliness and poor attachment become more prominent, stimulating grieving, which for some people can seem overwhelming and result in blocking. For example, one recent patient acknowledged that as she began to experience compassion for herself she became aware of how lonely she had been as a child, and fearful of her parents. "All my life I seem to have felt cut off from others, really." Compassion is a process which can start to facilitate exposure to these avoided feelings and memories but in doing so, grieving and sadness are often part of the process. Recently, Lecours and Bouchard [59] have shown that borderline personality disorder severity is linked to difficulties in experiencing sadness and grief.
The capacity to develop compassion for self and others probably depends upon some capacity for empathy and mentalizing, which for some people can be compromised especially when under stress [60]. So it is plausible that difficulties with empathy and mentalising will also compromise compassion [52].
Another major reason people can run into difficulties with compassion and affiliative positive emotion is partly because the selfmonitoring and self-correcting systems have been primarily entrained to self-criticism rather than self-validation or self-support. Selfcriticism is of course the opposite of self-compassion, especially when associated with feelings of anger and contempt for the self [42,43,61,62]. Self-criticism has long been linked to increased vulnerability to psychopathology [63] and poor recovery [64]. There is an important link between fears of compassion and internal feelings of hostility towards the self and self-criticism [51,52]. Self-critical individuals often struggle with standard therapies [64,65] and have become a focus for compassion focused therapy [13,14]. A series of studies have shown that shame-prone and self-critical individuals tend to come from more difficult attachment backgrounds and have problematic attachment strategies which can interfere with being open to affiliative and compassionate feelings [34].
Indeed, self-critics can respond to compassionate imagery with threat responses as measured by reduced heart rate variability [66], greater amygdala activation [67] and are less likely to have pleasurable affiliative experiences to compassion imagery when given oxytocin [68]. In fact, as noted above, when self-critical people are invited to focus on feeling connected and being open to receiving compassion, care and concern from others, this often stimulates feelings of grief or an awareness of "how lonely one feels", and/or a sense of not deserving compassion [68]. In addition, research on emotion processing has shown that individuals who are depressed, anxious, insecurely attached or self-critical show diminished emotional processing of affiliative, kind and compassionate social cues such as positive facial expressions [69][70][71].
So this review shows that compassion is complex, it can be focused on others, self or receiving compassion and there are a range of reasons why individuals may block compassion. Typically, receiving compassion, particularly in the context of distress, is especially difficult for some people. However, affiliative emotions, and the capacity to be emotionally regulated through caring and affiliative relationships are central to human functioning and recovery from depression. Patients who can't access these evolved systems, because of fear and avoidance, are cut off from major sources of affect regulation and the contexts for developing social cognition [36]. Hence this area requires further research on affiliative emotion in clinical populations.

Aims
This study explores the meta-cognitive fears of compassion in a clinical population and their link to self-criticism, self-compassion and the psychopathology variables of depression, anxiety and stress. We hypothesise that, as in nonclinical populations, in clinical populations, fears of compassion will be highly, negatively correlated with abilities to be self-compassionate, positively correlated with self-criticism and self-coldness, and significantly associated with depression, anxiety and stress.

Participants
Depressed patients (N=53) from the acute wards (N=12) and day hospitals (N=20) of the Derbyshire Healthcare NHS Foundation Trust, Leicestershire Mental Health Trust (N=4) and from the local Derbyshire self-help depression support group (N=17) participated in the study. Participants were 32 women and 21 men, ages ranged from 26 to 91 years (M=50.36; SD=11.38). Given the heterogeneity of this group they were subject to a range of interventions including medication, individual and group therapies. We are not exploring the impact of particular therapies on fears of compassion however. All participants completed a series of self-report scales.
Although no research diagnostic interview was given, the criteria for inclusion were having a diagnosis of a depressive disorder and being treated for depression. Eight patients in the original sample had scored within the 'normal' range (scores of 0-9) of the DASS cut-off scores for depression and so their data were excluded from the analyses which left 53 people in the final sample. According to the DASS cut-off scores, 7.6% of patients were classified as mild; 13.2% were moderate; 13.2% were severe and 66% were extremely severe in terms of their depression scores. Ethical approval for the study was received from the Derbyshire NHS Ethics Committee. All participants gave informed consent.

Fear of compassion scales
This study used the recently developed fears of compassion scales (Gilbert, McEwan, Matos et al., 2011). Fear of compassion for Self scale comprises 15 items (e.g. "Getting on in life is about being tough rather than compassionate"); fear of compassion from Others scale comprises 13 items (e.g. "Wanting others to be kind to oneself is a weakness"); fear of compassion for Others scale comprises 10 items (e.g. "I fear that being too compassionate makes people an easy target"). The items were rated on a five-point Likert scale (0=don't agree at all, 4=completely agree). These scales showed good reliability with Cronbach's alpha's of .92 for self, .85 from others, and .84 for others in a student sample.
for this study to avoid confusion with attachment issues. Respondents are asked to rate how true each compassionate statement is on a sevenpoint Likert scale ranging from 1 (not at all true of me) to 7 (very true of me). This scale has been found to have a good Cronbach's alpha value of 0.95.

Self-compassion scale (SCS)
This 26-item scale assesses levels of self-compassion [44]. There are three factors of positive self-compassion: Self-kindness, Common humanity and Mindfulness; and three factors that focus on a lack of self-compassion and negative self-evaluation: Self-judgement, Isolation and Over-identification. We obtained two totals for this scale: Self-compassion (sum of the three positive factors) and Selfcoldness (sum of the three negative factors). Participants indicate how often they engage in these ways of self-relating on a Likert scale 1 -5. The scale has good reliability (Cronbach's alphas ranging from .75 to . 81). Note: In our view it is important to analyse the scale as two separate constructs because the self-coldness factors are highly correlated with self-criticism and shame. Therefore correlations between psychopathology and self-compassion could simply be due to the known link between shame, self-criticism and psychopathology. Also research suggests that self-compassion/reassurance operates through different physiological systems than self-criticism [67].

Forms of self-criticism/Self-reassuring scale (FSCRS)
This 22-item scale assesses participants' thoughts and feelings about themselves during a perceived failure [61]. Two subscales measure forms of self-criticising (Inadequate self and Hated self) and one subscale measures tendencies to be reassuring to the self (Reassured self). Participants respond on a Likert scale (0=Not at all like me, 4=extremely like me). The scale has good reliability with Cronbach's alphas of .90 for Inadequate self, .86 for Hated self, and 0.86 for Reassured self.

Depression, anxiety and stress scale (DASS-21)
This 21-item shortened version of the DASS-42 consists of three subscales measuring Depression, Anxiety and Stress [73]. Participants are asked to rate how much each statement applied to them over the past week, on a Likert scale 0-4. The DASS-21 subscales have Cronbach's alphas of .94 for Depression, .87 for Anxiety and .91 for Stress [74].

Data analysis
Analyses were conducted using SPSS version 18 [75]. Total scores were calculated for each of the subscales in the analysis. The data were checked for normality of distribution and outliers using scatterplots. Skewness values ranged from -0.09 to -1.56 and kurtosis values ranged from 0.11 to 3.15. Inadequate self-scores were slightly skewed and kurtotic due to high scores, hence this variable was log transformed for further analysis.

Descriptive analysis
The means, standard deviations and Cronbach's alphas of the variables studied are shown in Table 1. Note: FOC, Fear of Compassion; FSCRS, Forms of Self Criticising/ Attacking and Self Reassuring Scale; DASS, Depression, Anxiety, and Stress Scale.

Compas sion for Others
Gender differences were explored in the self-report scores (21 males, 32 females) with an independent measures t test. Results showed a significant difference for the Stress subscale of the DASS [t (51)=-2.14, p=.037], with females mean scores being significantly higher (M=15.88, SD=3.75) than males (M=13.05, SD=5.89).

Correlation analysis
Pearson's Correlation Coefficients (two-tailed) were conducted for fears of compassion and all the study variables (Table 1). All correlations are displayed for transparency.

Fear of compassion for others
As can be seen in Table 1, fear of giving compassion to others was only moderately positively correlated with fear of receiving compassion from others and anxiety. Fear of compassion for others had no correlation at all with feeling compassion for others as measured by the Compassionate Love scale [72].

Fear of compassion from others
Fear of being open and receptive to receiving compassion from others was highly positively correlated with fear of being compassionate to oneself (r=.75). Fear of receiving compassion from others shared negative correlations with self-compassion and selfreassurance and had positive correlations with self-coldness, selfcriticism (both inadequate self and hated self), depression, anxiety and stress. Of particular interest was the high correlation (r=.62) between fear of compassion from others and depression.

Fear of compassion for self
Fear of compassion for self is significantly positively correlated with self-criticism and self-coldness, depression, anxiety and stress. It is negatively correlated with self-compassion and self-reassurance. One variable with which fear of compassion for self-did not show any correlation was compassion for others.
Taken together, compassion for others seems to function differently than self-compassion and being open to receiving compassion from others

Regression
Based on the primary aim of this study, to explore how fears of compassion predict depression, a multiple linear regression analysis was conducted with fears of compassion from others and for self as the independent variables and depression as the dependent variable. Since fear of compassion to others does not correlate with any of the main variables it was not included in the regression. Analysis of regression plots of standardized residuals against standardized predicted values, histograms and P-P plots showed that assumptions of linearity and normality of distribution were met. For a more conservative analysis of results, given the small sample size, we bootstrapped the regression model. The analysis accounted for 42% of the variance in the prediction of depression [F (2,50)=17.89, p=.000]. Fear of compassion for self was non-significant (β=.33, p=.087) and fear of compassion from others approached significance (β=.36, p=.059). Fears of compassion for self and from others were highly correlated so multicolinearity was tested for in the regression model. There was no evidence of multicolinearity (i.e. tolerance>.20 & VIF<5.00) [75].
Taking into account the large correlations between fears of compassion from others and for self, and between inadequate self and hated self-criticism (all r's>.6) we summed these variables to produce total scores for fears of compassion and self-criticism and entered these into a regression to explore their contribution to depression. We did this with the rationale of simplifying the analyses by entering fewer variables into the prediction of depression. Again, assumptions of normality and linearity were met and the regression was bootstrapped for a more conservative analysis. Both fears of compassion (β=.36, p=. 018) and self-criticism (β=.44, p=.004) made significant contributions [F (2,50)=28.40, p=<.001] and accounted for a very high 53% of the variance.
We then conducted a mediator analysis to see if the relationship between self-criticism and depression was perhaps mediated by fears of compassion. Mediator analyses were conducted using multiple regressions, following the four-step analysis recommended by Baron and Kenny [76]. Self-criticism was entered as the predictor variable, fears of compassion as the mediator and depression as the dependent variable.
Step 1 found that the predictor variable self-criticism was a significant predictor of the dependent variable depression [F (1,51) Step 4 analysis of the standardised β weights indicates that fears of compassion partially mediates the relationship between self-criticism and depression. A Sobel test supported this (Sobel test=2.54, p>.01). That is, whilst self-criticism directly predicts depression, depression is especially high if, in addition, one scores higher in fears of compassion ( Figure 1).

Discussion
This study explored the fears of compassion in a depressed population. As in the nonclinical population [51], being fearful of compassion from others, is highly correlated with being fearful of being self-compassionate (r=.75). This is consistent with qualitative research reporting that depressed individuals find it difficult to imagine receiving compassion from others or being selfcompassionate especially when they are in the state of depression [16,[77][78][79]. So there seems to be a general block and fear of compassion, be it coming from other people or oneself.
A second core finding is that when people are self-critical they're not just hostile to themselves but they may be actively resisting experiencing positive and affiliative emotions -as shown by the high correlations between fears of compassion and self-criticism. Indeed, fears of receiving compassion from self or from others are highly inversely correlated with self-reassurance and the warmth dimension of the self-compassion scale. Even if one is able to reduce self-criticism in therapy this does not automatically guarantee that people will be able to be open to positive caring signals and stimuli in the form of affiliative relating.
The fears of receiving compassion from others and being selfcompassionate are highly correlated with our psychopathology measures as is self-criticism. We also explored the relative contribution of fears of compassion (fear of compassion from others and for self) and self-criticism (inadequate self and hated selfsummed) to depression, with a multiple regression. Both the fears of compassion and self-criticism were significantly linked to depression and made independent contributions in a mediator analysis. This highlights again that the concepts of positive emotions associated with self-compassion/kindness and self-reassurance should be seen as separate constructs from negative, critical and condemning dimensions of self-evaluation as they have distinct effects and are related to two very different (neurophysiological) processing systems [67].
There are a number of implications of this research for the development of psychotherapy for depression. For example, further research is needed to understand how these fears operate, how they may vary from person to person, therefore requiring different therapeutic interventions. There is qualitative data to suggest that although depressed patients understand the value of compassion and would like to develop it for themselves, they fear compassion and feel they don't deserve it and are unable to generate or receive it [77]. It could be argued that fears of compassion and affiliative emotions could be treated therapeutically in much the same as any other emotion (e.g. anxiety) such as gaining insight, desensitisation, gradual exposure and developing capacities to tolerate and feel the emotion [14]. When conducted therapeutically, compassion interventions seem to overcome these fears [16]. In a recent study, Jazaieri et al. [26] developed compassion cultivation training based on Buddhist concepts and showed significant reductions in the fears of compassion with this training. CFT has also been shown to significantly reduce fears of compassion and for it to become a therapeutic process in series of steps [80]. So we suggest that clinicians should be aware of the complexities of stimulating and cultivating affiliative and compassion focused emotions but also appreciate the importance of doing so and the significant effects of affiliation on affect regulation and possible impacts on depression [36].

Limitations
There are limitations to this early research into fears of compassion in a depressed population. One is the relatively small sample size, indeed recruitment from mental health wards was slow due to the complex and comorbid nature of patients being treated on in-patient wards. Attempts were made to recruit further participants from day hospital and self-help groups. However, 8 of these had to be removed from the final analysis as they did not meet to clinical cut-off scores for the DASS. A challenge of focusing on a specific mental health problem such as depression, rather than a general psychiatric sample is that patients in psychiatric services frequently have comorbidities and mixed diagnoses and these can influence responding in questionnaires. Nonetheless we believe that as a first study it highlights the importance of studying the role of affiliative emotion and compassion cultivation in depression in new ways.