Introduction

According to evolutionary and neurosciences research, the motivations and behaviours of humans may be understood through an emotion regulation model composed of three interacting systems [1,2,3]. These affect regulation systems which have been defined as: threat-protection, drive-resource seeking, and safeness-soothing systems [1, 3, 4]. The threat system is focused on detection and protection from threats, through the rapid activation of behavioural (like fighting, escaping, or freezing) and emotional defensive responses (such as anger, anxiety, or disgust) [2]. This system can be triggered by diverse threat-signalling stimuli, such as external (e.g., physical attacks or threatening social interactions) and internal events (e.g., feelings of inferiority and shame) [2, 3]. The drive system involves positive feelings, which energize and motivate individuals to seek out for the needed resources and important life goals [1]. The soothing system, associated with the attachment system, involves a different type of positive feelings linked to affection, kindness, and sense of peace and well-being [3]. According to Gilbert [4], this soothing-safeness system can reduce distress in response to threats and down-regulate reward-focused behaviours, promoting a feeling of contentment, connectedness, and well-being.

Embedded in the soothing system is compassion, described by Gilbert [2, 5] as a sensitivity and awareness for one’s and others’ suffering, as well as a courage to adopt attitudes to alleviate or prevent painful experiences. Two processes are inherent in this definition of compassion: (1) the ability to face and deal with suffering, instead of avoiding or ignoring it; and (2) the readiness to develop the necessary skills to alleviate or prevent suffering and painful events [6]. Several accounts suggest that compassion can flow in three different ways: we can receive compassion from others, express compassion attitudes to others, as well as to ourselves [3, 5, 7]. Growing evidence has shown that developing compassion is associated with an increased positive affect and quality of life, namely social, physical, and emotional well-being [8,9,10].

Although these affiliative emotions are associated with positive feelings and well-being, for some people, compassionate emotions are experienced as frightening and may be linked with avoidance or fear reactions [11]. These fears can be understood in light of the attachment model. In fact, the literature has shown that individuals with a secure attachment tend to perceive others as sources of support, acceptance, and care, while individuals from insecure backgrounds tend to present distress with interpersonal closeness [12, 13]. In this sense, for insecure individuals, signals of compassion and kindness may reactivate negative affiliative memories and trigger fears of compassion [14, 15].

Increasing evidence suggest that fears of compassion are associated with maladaptive emotional regulation mechanisms (such as self-criticism and self-coldness) and psychopathological symptoms (e.g., depression, anxiety, and stress) [11, 16]. Even though the impact of fears of affiliative emotions on eating psychopathology remains largely unexplored, the previous findings offered evidence that fears of compassion are positively associated with body image and eating difficulties [17, 18]. Particularly, Kelly and colleagues [17] demonstrated that eating disorder patients who reported lower self-compassion and higher levels of fears of compassion presented more severe symptomatology at the beginning of the intervention, and lower results in a 12-week program for eating disorders.

Fears of compassionate and affiliative emotions may have a negative impact on social relationships, due to difficulties in experiencing feelings of connectedness and safeness in social interactions [19]. Kelly and colleagues [20] suggested that social safeness, characterized by the presence of feelings of warmth, acceptance, and connectedness within interpersonal interactions, is negatively associated with different maladaptive emotional processes and psychopathological indicators (such as inferiority feelings and shame) [19].

Shame is a painful emotion, conceptualized as a defensive response that arises when one thinks that others see and judge the self negatively, as weak, unattractive, inferior, and/or defective—external shame [21, 22]. However, these feelings can be internalized leading to negative self-evaluations and feelings of inadequacy and inferiority—internal shame [21]. These negative and shaming evaluations can be focused on weight and physical appearance. Body image-focused shame involves the perception that one has physical attributes or characteristics (body shape, size, or weight) that fail to fit within what others perceives as attractive [23]. Indeed, body shame is highly dependent of the sociocultural context, since it is cultural context that decides what is valued or discriminated [22, 24]. The growing objectification of body by society, particularly in relation to female appearance, leads women to set weight and physical appearance-related high and unrealistic standards, contributing for a negative self-evaluation and higher levels of body shame [25, 26]. Although body-focused shame seems to have a central role in eating psychopathology [23, 26, 27], the role of body shame on disordered eating remains poorly documented.

There is evidence that dietary restrain may be seen as a strategy to reach a body image closer to ideal body image patterns, that emerges to overcome social difficulties and to compete for positive social attention [28, 29]. In fact, the literature had shown that drive for thinness can be conceptualized as a strategy of insecure striving, particularly in women who experience feelings of inferiority and shame, based on their body image [28, 30, 31].

The present study aimed to test a model that explores the impact of different dimensions of fears of compassion on disordered eating. Furthermore, this model examined whether the mechanisms of social safeness and body shame mediate these relationships. It is expected that higher levels of fears of compassion would be associated with lower levels of social safeness and connectedness and higher levels of body shame, which may give rise to higher levels of disordered eating.

Materials and methods

Participants

The sample was composed of 645 female participants, aged between 18 and 55 (M = 27.74; SD = 8.76), with a mean of 14.78 (SD = 2.52) years of education. Concerning BMI, 41 (6.4%) participants were underweight (BMI < 18.5 kg/m2), 435 (67.4%) presented a normal weight (18.5 kg/m2 ≤ BMI ≤ 25 kg/m2), and 169 (26.2%) were overweight (25 kg/m2 < BMI), according to the WHO [32], reflecting the normal distribution for female adults in the Portuguese population [33]. It is important to note that 5.58% (n = 36) of the current female sample revealed severe eating psychopathology symptomatology (EDE-Q > 4; [34]).

Please see Table 1 for a more detailed characterization of the sample.

Table 1 Sample’s demographic (N = 645)
Full size table

Measures

The research battery included demographic data and the Portuguese versions of the following self-report measures:

Body Mass Index (BMI included demographic data). Participants’ BMI was calculated through the Quetelet Index based on self- reported current weight and height (Kg/m2).

Fears of Compassion Scales (FCS; [11]; Portuguese version by Simões & Pinto-Gouveia 2012). FCS is a self-report instrument composed of three subscales: (1) Fear of compassion for self (FCS_ForSelf), with 15 items that assess the fear and resistance of demonstrating affiliative emotions and compassion for the self (e.g., “I worry that if I start to develop compassion for myself I will become dependent on it”); (2) Fear of compassion from others (FCS_FromOthers), with 13 items, that measures fears of receiving kindness and compassion from others (e.g., “If I think someone is being kind and caring towards me, I ‘put up a barrier”); (3) Fear of compassion for others (FCS_ForOthers), with 10 items that rates the fear of expressing sensitivity or compassion for others (e.g., “I worry that if I am compassionate, vulnerable people can be drawn to me and drain my emotional resources”). Participants were asked to rate on a five-point scale (0 = “Don’t agree at all”–4 = “Completely agree”) and their agreement with each item. These scales revealed good reliability, in the original version (α = 0.92, 0.85, and 0.84 for FCS_ForSelf, FCS_FromOthers, and FCS_ForOthers, respectively; [11]), as well as in the Portuguese version (α = 0.94, 0.91; and 0.88, respectively).

Social Safeness and Pleasure Scale (SSPS; [19]; Portuguese version by Pinto-Gouveia, Matos, & Dinis, 2008). SSPS is a self-report instrument, composed of 11 items, designed to assess individual’s sense of warmth, safeness, and reassurance in their social relationships. Participants were asked to rate the items on a five-point scale (from 0 = “Almost never” to 4 = “Almost all the time”). SSPS has revealed excellent psychometric properties, particularly strong construct and discriminant validity and good reliability (α = 0.92; [19]) and a unidimensional structure and good psychometric characteristics in the Portuguese validation study.

Body Image Shame Scale (BISS; [23]) BISS is a 14-item self-report measure that assesses body image-related shame, considering its internal (e.g., “It bothers me to see my body undressed”) and external (e.g., “I feel uncomfortable in social situations, because I feel that people may criticize me because of my body shape”) dimensions. Using a five-point scale (from 0 = “Never” to 4 = “Almost always”) respondents are asked to rate each item. The instrument showed good psychometric properties, with alphas of 0.92 for global scale, 0.89 and 0.90, for Externalized and Internalized Body Shame [23].

Eating Disorder Examination Questionnaire (EDE-Q; [35]; Portuguese version by [36]). The EDE-Q is a 36-item self-report inventory that assesses disordered eating attitudes and behaviours. This questionnaire comprises four subscales: Restraint, Shape Concern, Weight Concern, and Eating Concern. The sum of the average of each subscale score offers a global measure of eating psychopathology. The items are rated for their frequency or severity on a scale from 0 to 6. This instrument showed good psychometric properties, demonstrating a α ≥ 0.94 in different study samples, for both the original [35] and the Portuguese versions [36].

The internal reliability coefficients of each measure used in this study are reported in Table 2.

Table 2 Means (M), standard deviations (SD), Cronbach’s alphas (α), and intercorrelation scores on self-report measures (N = 645)
Full size table

Procedure

The present study is part of a wider research project about eating-related attitudes and behaviours and emotion regulation processes in the Portuguese population. All ethical and deontological requirements inherent to scientific research were respected: the Ethic Committees of the institution involved in the investigation approved the research protocol, and all participants provided their inform consent after being fully informed about the nature and objectives of the study, the voluntary nature of their participation and the confidentiality of the data, which was only used for research purposes. Participants were recruited through online advertisements, using a social network (Facebook) in which an Internet link for an online version of the questionnaires was included. Participants were also collected through private messages on Facebook and were asked to share it with two more friends (Exponential Non-Discriminative Snowball Sampling method). Data were collected through LimeSurvey, a survey tool that ensures the confidentiality of data. In accordance with the aims of this study, data were cleaned to exclude participants who: (1) do not have Portuguese nationality and (2) were younger than 18 and older than 55. This process resulted in the final sample of 645 participants.

Analytic strategy

Data analysis and path analysis were performed using the software IBM SPSS Statistics 22.0 [37] and AMOS [38] (Version, 23).

Preliminary data analyses were performed to test the univariate and multivariate normality of our sample through the calculation of Skewness and kurtosis coefficients.

To analyse the sample characteristics, descriptive statistics (means and standard deviations) were performed. Furthermore, Pearson’s correlation coefficients were conducted to explore the associations between BMI, fear of compassion for others, fear of self-compassion, fear of compassion from others, social safeness, body shame, and disordered eating. Following Cohen’s guidelines, the correlations magnitudes were considered weak (between 0.1 and 0.3), moderate (between 0.3 and 0.5), and strong (above 0.5), at a significance level of 0.05 [39]. In addition, a path analysis was run to test the presumed relations proposed by the theoretical model under study. Specifically, we explored the relation between fears of compassion (exogenous variables) and disordered eating (endogenous variable), mediated by social safeness and body shame (mediator variables), controlling for the effect of BMI.

To test the regression coefficient’s significance and fit statistics, a maximum-likelihood estimation method was used. A series of goodness-of-fit indices (e.g., Chi-square, CMIN/DF, TLI, CFI, and RMSEA) were performed to test the model consistency [40]. Bootstrap re-sampling was further used to test the significance of the mediation paths, using 5000 bootstrap samples and 95% confidence intervals (CIs; [42]).

Results

Preliminary data analyses

Results of skewness—which ranged from − 0.46 (SSPS) to 1.56 (BMI)—and Kurtosis—which varied from − 0.41 (FCS_Fears of Compassion For Others) to 3.38 (BMI)—indicated no severe violation of normal distribution [41].

Descriptive and correlations analyses

The means, standard deviations, and correlations between study variables are presented in Table 2. Results indicated that the three dimensions of fears of compassion presented moderate-to-high positive correlations with each other. Furthermore, fear of compassion for others, for self, and from others were negatively associated with social safeness (with low, moderate, and high magnitudes of correlations, respectively). Positive small-to-moderate associations were verified between fears of affiliative and compassionate feelings and body-related shame. In addition, body shame was negatively linked with feelings of connectedness and social safeness. Moreover, disordered eating behaviours were positively and moderately associated with all dimensions of fears of compassion and highly linked with body-related shame.

Finally, BMI revealed a small association with higher levels of fears of compassion and lower levels of social safeness, and a moderate-to-high associations with body shame and disordered eating.

Path analyses

The path analysis was performed to test whether the mechanisms of social safeness (SSPS) and body shame (BISS) mediate the impact of the three dimensions of fears of compassion (FCS) on disordered eating, while controlling for the effect of BMI.

Initially, the path model was tested through a fully saturated model (i.e., with zero degrees of freedom), consisting of 35 parameters, which explained 29% of social safeness, 42% of body shame, and 62% of disordered eating variance. The following paths were not significant: the direct effect of fear of receiving compassion from others on disordered eating (bFCS_Fears of Compassion From Others = 0.00; SEb = 0.01; Z = − 0.15; p = 0.88), the direct effect of fear of expressing compassion for others on body shame (bFCS_Fears of Compassion For Others = − 0.00; SEb = 0.00; Z = − 0.34; p = 0.74), the direct effect of fear of self-compassion on disordered eating (bFCS_Fears of Compassion For Self = 0.00; SEb = 0.00; Z = 0.52; p = 0.60), the direct effect of social safeness’ direct effect on disordered eating (bsocial safeness = 0.00; SEb = 0.00; Z = 0.51; p = 0.61), and the direct effect of fear of expressing compassion for others on social safeness (bFCS_Fears of Compassion For Others = 0.03; SEb = 0.04; Z = 0.79; p = 0.43). These paths were progressively removed and the readjusted model was then tested.

The final adjusted model (Fig. 1) explained 29, 42, and 62% of social safeness (SSPS), body shame (BISS), and disordered eating (EDE-Q) variances, respectively. The model revealed an excellent model fit [X2(6) = 1.82; p = 0.94; CMIN/DF = 0.30; TLI = 1.01; CFI = 1.00; RMSEA = 0.00; p = 1.00; 95% CI = 0.00–0.01].

Fig. 1
figure 1

Final path model. Standardized path coefficients among variables are presented. All path coefficients are significant at the 0.05 level; ***p < 0.001

Full size image

All paths coefficients were statistically significant (p < 0.05). Specifically, fears of compassion for others presented a direct effect of 0.11 (bFCS_Fears of Compassion For Others = 0.02; SEb = 0.00; Z = 4.31; p < 0.001) on EDE-Q. Furthermore, fears of compassion for self presented a direct effect of − 0.14 (bFCS_Fears of Compassion For Self = − 0.12; SEb = 0.04; Z = − 3.08; p = 0.002) on social safeness and of 0.17 (bFCS_Fears of Compassion For Self = 0.01; SEb = 0.00; Z = 4.04; p < 0.001) on body image shame. Furthermore, fears of compassion from others showed a direct effect of − 0.43 (bFCS_Fears of Compassion From Others = − 0.37; SEb = 0.04; Z = − 9.22; p < 0.001) on social safeness. In turn, fears of compassion from others and social safeness presented a direct effect on body image shame of 0.19 (bFCS_Fears of Compassion From Others = 0.02; SEb = 0.00; Z = 4.31; p < 0.001) and of − 0.24 (bsocial safeness = − 0.02; SEb = 0.00; Z = − 6.65; p < 0.001), respectively. Finally, body image shame demonstrated a direct effect of 0.64 (bbody image shame = 0.85; SEb = 0.04; Z = 23.18; p < 0.001) on EDE-Q.

The analysis of indirect effects revealed that fear of compassion for self had an indirect effect of 0.03 (95% CI = 0.01–0.06) on body image shame, which was partially mediated through the effect of social safeness. Fears of compassion for self also presented an indirect effect on EDE-Q of 0.13 (95% CI = 0.06–0.20), through the mechanisms of social safeness and body shame. Furthermore, fears of compassion from others showed an indirect effect of 0.10 (95% CI = 0.06–0.15) on body image shame, which was partially mediated by social safeness. In turn, fears of compassion from others also demonstrated an indirect effect on EDE-Q of 0.19 (95% CI = 0.12–0.25), through the mechanisms of social safeness and body image shame. Finally, social safeness had an indirect effect of − 0.15 (95% CI = − 0.20 to − 0.10) on EDE-Q, which was totally mediated by body image shame.

To sum up, the final model accounted for 62% of EDE-Q’s variance, demonstrating that fear of expressing compassion for others directly impacts on EDE-Q, and that fear of self-compassion and fear of receiving compassion from others partially impact on EDE-Q through the mechanisms of social safeness and body shame.

Discussion and Conclusion

The current evidence suggests that fears of compassion are linked with several emotional regulation difficulties and psychopathological conditions [11, 16]. However, the impact of these fears of affiliative emotions on disordered eating has not been clearly studied. The present study aimed at examining the associations between fears of compassion, social safeness, body-related shame, and disordered eating. The tested model hypothesized that the connection and proximity with others and feelings of devaluation based on physical appearance would mediate the impact of different dimensions of fears of compassion on disordered eating attitudes and behaviours, when controlling for BMI.

Correlation analyses results supported these hypotheses and add to the current knowledge revealing the significant relationships between the three dimensions of fears of compassion and social safeness, body shame, and disordered eating. Moreover, comparing to fears of expressing compassion for others, fear of being self-compassionate, and of receiving compassion from others presented stronger associations with the remaining study variables. These findings are in agreement with previous data which postulated that these dimensions of fears of compassion are strongly linked with maladaptive emotional processes and different psychopathological indicators, when compared to fear of being compassionate to others [11]. Furthermore, the current data are in line with and extended prior research, demonstrating that fears of affiliative emotions (specially fear of self-compassion and fear of receiving compassion) were negatively associated with one’s ability to feel secure and connected to others [3, 19]. Fears of being self-compassionate and of receiving compassion were demonstrated to be linked with feelings of inferiority based on weight and body image and with the adoption of maladaptive eating-related behaviours [17, 18]. The present study also found a negative correlation between social safeness and body shame, corroborating that feelings of insecurity in social relationships are associated with greater feelings of inferiority, as suggested by Gilbert [3]. Finally, the strong positive association between body image and disordered eating was corroborated [28, 30].

To further examine these associations, a path analysis was run to test the mediating effect of social safeness and body shame on the relationship between fears of compassion and disordered eating. Results indicated that the theoretical model is plausible, and that it accounted for 62% of disordered eating’s variance. Consistent with prior evidence [11], results confirmed that the three dimensions of fears of compassion are positively associated with each other. However, the different dimensions of fears of compassion presented distinctive effects on study variables. It is noteworthy to mention that fear of expressing compassion for others revealed a direct effect on disordered eating. This is an interesting result that merits greater consideration in future studies. Although, in this model, the fear and resistance of demonstrating affiliative and compassionate behaviours for others have presented a direct effect on disordered eating, future studies should explore the presence of other emotional processes that function as mediators of this association. This goes in line with the previous studies which suggested that fear of expressing compassion for others may act through different mechanisms than the other two dimensions of fears of compassion [11].

In contrast, our results revealed that fear of self-compassion and of receiving compassion from others presented an indirect effect on disordered eating, through the mechanisms of social safeness and body shame. These findings fit with recent research on how fears of compassion fuel negative emotions and undermine mental well-being, and suggest that distinct forms of fears of expressing and receiving affiliative emotions may impact on body image difficulties. Moreover, the mediational paths uncovered in the current study suggest that the negative impact of fearing affiliative and compassionate feelings on eating behaviour depends on the extent to which these fears contribute to greater difficulties in close social interactions and higher levels of inferiority and unattractiveness based on body image. These results are aligned with the previous literature. Body image is regarded as a key domain for self and social evaluation (that is to estimate whether one is valued and accepted or criticized or even rejected by their social group), especially among women [e.g., 30]. In this way, and according to Duarte and colleagues [23], the painful emotion of body image shame arises in a context of concerns with belonging and acceptance by others. This emotion and associated defensive behaviours may increase the scrutiny and monitorization of physical appearance and the adoption of a rigid control of eating behaviour as a possible solution to avoid these shameful and aversive experiences and to seek. The present study extends the current knowledge on this area by exploring how fears of expressing compassion are associated with social safeness and body image and disordered eating, i.e., body image shame and disordered eating, respectively. In summary, our findings allow the hypothetization that fear or resisting to express compassion for the self and others is associated with greater insecurity within social relationships and greater feelings of body image shame, which may undermine one’s adoption of healthy eating behaviours. Thus, the findings of this study suggest that women who present higher levels of fear of self-compassion and of receiving signs of affection, empathy, and compassion from others tend to feel more insecure and less close within their social group, which seems to explain a more severe experience of body shame and the consequent adoption of disordered eating behaviours.

Although these are promising findings, this study presents some limitations. First, its cross-sectional nature does not allow presuming causal conclusions. Thus, experimental and longitudinal designs are warranted to clarify the associations between these variables. Furthermore, the use of self-report measures and the recruitment through an online survey may not allow the generalization of data. This recruitment method does not allow assessing participants’ clinical criteria for psychiatric disorders, which may play an important part in the studied model. In addition, self-reported height and weight may not be accurate, which can lead to a biased BMI. Therefore, future studies should include other assessment methods. Furthermore, even though body image and eating difficulties are more prevalent in women, men also experience these symptoms; upcoming studies should thus focus on men samples to further support the adequacy of this model. Finally, although the main aim of the current study was to specifically address the mediator role of social safeness and body shame on the link between fears of compassion and disordered eating, future research should expand this model by considering other relevant variables that may contribute to this association.

Nonetheless, this is the first study that clearly demonstrates the pervasive role of fears of affiliative and compassion experiences on disordered eating via decreased social safeness and increased body shame. These data seem to support the adequacy of compassion-focused approaches focused on overcoming of fears of affiliative experiences to promote healthy body image attitudes and eating behaviours.