Canadian Journal of Behavioural Science / Revue canadienne des sciences du comportement ISSN: 0008-400X © 2019 Canadian Psychological Association 2020, Vol. 52, No. 2, 159 –170 http://dx.doi.org/10.1037/cbs0000150 Is Self-Compassion Contagious? An Examination of Whether Hearing a Display of Self-Compassion Impacts Self-Compassion in the Listener Kathryn Miller and Allison Kelly University of Waterloo Self-compassion is associated with various emotional and physical benefits, making it important to investigate the social contexts that facilitate it. This study was the first to examine whether exposure to another person’s display of self-compassion affects the listener’s state level of self-compassion. One hundred and 19 female university students listened to a self-compassionate, self-esteem enhancing, or neutral reaction to an academic failure; wrote about their failure; and then completed measures of state self-compassion and affect. Results showed that after recalling a personal academic failure, hearing a self-compassionate reaction caused higher state self-compassion based on self-report and coded written responses, and lower negative affect than hearing a factual reaction whereas hearing a self-esteem enhancing reaction did not. Hearing a self-compassionate reaction did not yield significantly different outcomes in comparison to hearing a self-esteem enhancing reaction. These findings are the first to suggest that self-compassionate behaviour in others may influence how compassionately one feels and behaves toward oneself. Public Significance Statement Self-compassion is linked to greater resilience in the face of distress or failure, however, little is known about the social factors that may facilitate it. This study examined whether hearing someone else being self-compassionate facilitates one’s own self-compassion in the face of personal failure. Results support this idea such that hearing self-compassion from another seems to result in more self-compassionate attitudes in the listener. Keywords: self-compassion, self-esteem, social interactions, emotional contagion, coping Common humanity versus isolation is the ability to feel a sense of kinship in one’s hardship, knowing that everyone struggles, rather than feeling isolated in one’s hardship. Finally, mindfulness versus overidentification is the ability to view one’s distress in an objective way rather than being overcome by emotion (Neff, 2003). Self-compassion and self-esteem are both forms of positive self-regard and correlate moderately with one another (Neff, 2003); however, they are distinguishable constructs. Self-esteem refers to how much someone values himself/herself, (Crocker & Wolfe, 2001; Rosenberg, 1965), whereas self-compassion has no evaluative component; it refers to how much one cares for oneself and may benefit individuals most when self-esteem is low (Leary, Tate, Adams, Allen, & Hancock, 2007). Furthermore, numerous studies show that self-compassion contributes unique variance to markers of psychological well-being over and above self-esteem (Barnard & Curry, 2011). Unlike self-esteem, self-compassion is not correlated with narcissism, higher defensiveness, and increased aggression in the face of ego threat (Baumeister, Smart, & Boden, 1996; Leary et al., 2007; Neff, 2003). Self-compassion is a construct that has received much attention in recent years from social and clinical psychologists alike. Fundamentally, self-compassion is the ability to extend oneself care and compassion in the face of personal failure or distress (Neff, 2003). Just as one might soothe and support a close friend who is struggling, self-compassion involves approaching oneself with this same warm and reassuring mindset in times of hardship (Gilbert, 2005). Neff (2016; Neff et al., 2019) recently found that the most widely used measure of self-compassion, the Self-Compassion Scale (Neff, 2003), comprises two factors—a general self-compassion factor, with a secondary group factor that represents the six subscales or components of selfcompassion. These six components are composed of the presence of a positive attribute and the absence of a negative attribute. Self-kindness versus self-judgment represents the ability to treat oneself with kindness rather than criticism. This article was published Online First October 14, 2019. X Kathryn Miller and Allison Kelly, Department of Psychology, University of Waterloo. Correspondence concerning this article should be addressed to Kathryn Miller, Department of Psychology, University of Waterloo, 200 University Avenue West, Waterloo, ON N2L3G1, Canada. E-mail: k24mille@uwaterloo.ca Self-Compassion Levels as Varying and Malleable Although self-compassion shows a high degree of stability over time (Neff, 2003), there is also evidence that levels of self159 MILLER AND KELLY 160 compassion can fluctuate within a person over time. Experimental studies show that it is possible to induce a self-compassionate mindset and that doing so attenuates negative affect (NA) and preserves positive affect (PA; Adams & Leary, 2007; Johnson & O’Brien, 2013; Leary et al., 2007). More relevant to everyday life are findings from daily diary studies showing that 37– 42% of the variance in college students’ self-compassion levels occurred within persons (Breines, Toole, Tu, & Chen, 2014; Kelly & Stephen, 2016; Zuroff, Kelly, Leybman, Sadikaj, & Gilbert, 2012). Furthermore, these day-to-day changes in a person’s level of self-compassion were linked to corresponding adaptive changes in one’s body image, eating behaviour, and emotions (Breines et al., 2014; Kelly & Stephen, 2016; Zuroff et al., 2012). Evidently, vacillations in an individual’s level of self-compassion can meaningfully influence a person’s behaviour, feelings, and self-relating at a given time, making it important to identify the factors that contribute to fluctuations in self-compassion. Although levels of self-compassion appear to be malleable, little is known about what current situational factors might affect levels of self-compassion. From a practical standpoint, addressing this gap in the literature could help individuals understand what environmental contexts to seek out to inoculate themselves against negative self-attitudes, emotionality, and behaviours, and to experience greater well-being and social connectedness. In particular, it would seem quite possible that interpersonal processes and interactions in a given moment or period of time may influence an individual’s level of self-compassion in that moment or time period. To our knowledge, only one previous study has experimentally examined interpersonal influences on momentary levels of selfcompassion. Breines and Chen (2013) found that providing emotional support to someone led to greater levels of state selfcompassion when faced with a personal failure. Their study supports the general idea that one’s immediate social context may have an impact on one’s state level of self-compassion. An important next step is to understand what other interpersonal contexts may lead to increased self-compassion. We suggest that one such factor may be the level of self-compassion displayed by someone one observes and/or interacts with. Self-Compassion as Contagious? The idea that self-compassion levels in one person may influence self-compassion levels in another makes sense intuitively based on everyday social interactions. If a colleague is engaging in negative self-talk about their job performance, we may start to dwell on our own professional shortcomings. Conversely, witnessing someone who responds with self-compassion when they struggle at work may help us feel reassured and understanding toward our own career stressors. There is ample empirical evidence to support this idea that an intrapersonal process such as body image, willpower, and self-reinforcement in one person can affect that same process in others (Ackerman, Goldstein, Shapiro, & Bargh, 2009; Badaly, 2013; Kichler & Crowther, 2009; Marston, 1965; Oh & Damhorst, 2009), suggesting that self-compassion, too, might be “contagious.” Neff and Beretvas (2013) obtained findings consistent with this idea of self-compassion contagion, as they observed a positive correlation between the self-compassion levels of romantic part- ners. However, because this study was not experimental or longitudinal, it is impossible to know whether exposure to high selfcompassion in an interaction partner causes higher selfcompassion in oneself; that is, whether individuals simply selected partners with similar levels of self-compassion, or whether partners’ levels of self-compassion grew more similar over time, as in the proposed process of self-compassion contagion. There are at least two theories that might explain how selfcompassion may transfer from one individual to another. According to Social Learning Theory (Bandura, 1971), learning depends on experiencing or observing reinforcing consequences for performing a behaviour. Notably, this process can occur either consciously or unconsciously. (Hefferline, Keenan, & Harford, 1959; Hefferline & Keenan, 1963; Kennedy, 1970, 1971; Sasmor, 1966). It may be that when individuals treat themselves self-compassionately, the associated decrease in distress (Zessin, Dickhäuser, & Garbade, 2015) acts as vicarious reinforcement for an observer to do the same. Alternatively, this proposed interpersonal transmission of self-compassion could occur via simulation. Simulation Theory (Gallese, Fadiga, Fogassi, & Rizzolatti, 1996) proposes that humans automatically create neural representations of the mind-states, attitudes (i.e., self-compassion) and feelings of those around them (De Vignemont & Singer, 2006; Gallese, Eagle, & Migone, 2007; Wicker et al., 2003). Thus, when one observes self-compassion in someone else, consciously or unconsciously, the observer may automatically create an internal reproduction of a self-compassionate mindset and activate the neurological systems associated with self-compassion (e.g., caregiving motivational systems), thereby facilitating self-compassion. Finally, it may be that both social learning and mental simulation take place in tandem, such that observing a display of self-compassion leads to an internal simulation of self-compassion in the observer, and the experiential consequences of this simulated self-compassion (e.g., a reduction of personal distress) combined with the observation of its consequences in the other reinforce its subsequent use. Study Objectives The present study sought to test the novel idea that selfcompassion may be contagious across individuals. As a first step, we sought to examine the impact of hearing someone respond to a failure self-compassionately on participants’ state level of selfcompassion vis-à-vis a personal failure, using both self- and observer-ratings of self-compassion. We also wanted to examine the impact of hearing a self-compassionate display on affect, given that lowered NA and increased PA could be seen as “rewards” of self-compassion that could also be transferred from the speaker to the listener. In order to explore which outcomes were unique to hearing a display of self-compassion, and which were a more general effect of hearing someone display positive self-regard, we included a comparison condition in which participants heard someone describe a failure in a self-esteem-enhancing way. We also included a control condition in which someone spoke about a failure neutrally and factually. Because we were interested in whether self-compassion contagion can occur unconsciously, we opted not to draw the listener’s attention to the speaker’s coping style as described in more detail in the Method section. Our primary hypothesis was that after recalling a personal academic failure, hearing another person’s self-compassionate ac- IS SELF-COMPASSION CONTAGIOUS? count of their failure would yield higher state self-compassion in the listener than the other two conditions as measured by (Hypothesis 1a) observer ratings of self-compassion conveyed in participants’ written descriptions of their personal failures; and (Hypothesis 1b) self-reported self-compassion. As previous research has found that inducing a self-compassionate mindset when recalling a negative personal event results in lower NA than inducing a self-esteem enhancing mindset (Leary et al., 2007); Hypothesis 2 was that hearing a self-compassionate account of failure would produce lower NA in the listener compared to the self-esteem enhancing and neutral conditions. Hypothesis 3 was that hearing a self-compassionate or self-esteem-boosting description would yield higher PA than hearing a neutral description, given that both self-compassion and self-esteem are positive forms of self-relating, likely to increase PA. Finally, the effect of hearing a display of self-esteem enhancement on the listeners’ self-esteem levels was examined in an exploratory way to determine whether styles of self-relating that are similar to, but distinguishable from, selfcompassion may also be contagious. Method Overview of the Procedure The current study was composed of two online sessions completed 1 week apart. The first session consisted of self-report measures assessing the trait variables of self-compassion and selfesteem that would serve as covariates in certain analyses. A second online session was conducted separately, as we did not want participants to be primed to think about self-esteem and selfcompassion immediately before the manipulation, as we believed this might lead them to suspect the true objectives of the study. Furthermore, we wanted their postmanipulation reports of state self-compassion to reflect their momentary experience rather than simply being a carry-over effect from their trait self-report. During the second session, participants recalled a personal academic failure, and were then randomly assigned to listen to an audio clip of someone describing a personal academic failure in a selfcompassionate, self-esteem enhancing, or neutral way. Participants subsequently wrote about their personal academic failure and completed relevant outcome measures. Participants Female undergraduate university students were recruited via an online research participant pool at the University of Waterloo. Compensation was one credit toward a psychology course. Feelings of similarity and identification may influence emotional contagion (Goldstein & Cialdini, 2007), therefore the study was restricted to female participants to control for effects of similarity or identification that the participant might feel for the individual in the audio clip (i.e., women may automatically feel more similar to a woman in the audio clip and vice versa for men). This approach was more feasible as it allowed us to use one set of audio clips across participants, and required a smaller sample size. Research also shows that women are less self-compassionate than men (Yarnell et al., 2015), thus any increases in self-compassion that resulted from participation in this study may be more beneficial for women than men. 161 The initial recruited sample consisted of 210 participants, but of these 91 did not complete the second online session, which included the experimental manipulation. Although the attrition rate is somewhat high, it is not uncommon for two-part studies at this university, as students often reach their credit quota before completing both/all sessions of a study. Furthermore, as the participants were not assigned to an experimental condition until the second session, it can be assumed that attrition did not occur based on their assigned condition. The final sample consisted of 119 female undergraduates with a mean age of 20.66 years (SD ⫽ 2.68 years). Ethnic composition was 55 White (46.2%), 48 Asian (40.3%), two Middle Eastern (1.7%), three Black (2.5%), two West Indian (1.7%), six who indicated “other” ethnic group (5%), and three (2.5%) who declined to identify their ethnicity. Measures Trait measures. Trait self-compassion. The Self-Compassion Scale (SCS; Neff, 2003) is a 26-item scale measuring how compassionately people treat themselves in times of failure or distress. The scale assesses the three components of self-compassion—self-kindness versus self-judgment, common humanity versus isolation, and mindfulness versus overidentification. Items are thought to comprise six subscales, which reflect the positive and negative aspects of each component, and the mean of all items, with the negative ones reverse-scored, form a general self-compassion factor (Neff, 2003; Neff, Whittaker, & Karl, 2017). Items are rated on a Likert scale from 1 (Almost Never) to 5 (Almost Always). Items include “When I am going through a hard time I try to give myself the caring and tenderness I need.” The SCS has good convergent and discriminant validity and correlates with measures of psychological well-being such as higher life satisfaction and lower depression and anxiety (Neff, 2003; Neff, Kirkpatrick, & Rude, 2007; Neff & Vonk, 2009). Cronbach’s alpha was .94 in the current sample. Trait self-esteem. The Rosenberg Self-Esteem Scale (RSE; Rosenberg, 1965) measures how much one values or likes oneself. It is a 10-item Likert scale that measures dispositional self-esteem with items such as “On the whole I am satisfied with myself,” rated from 1 (Strongly Disagree) to 5 (Strongly Agree). Items ratings are summed to form a composite self-esteem score. The RSE demonstrates good internal reliability and validity (Rosenberg, 1965). Cronbach’s alpha was .93 in the present sample. State measures. Positive and negative affect. The Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) comprises two 10-item subscales that measured state PA and NA respectively. Items are rated on a Likert scale from 1 (Very Slightly) to 5 (Extremely). We administered the PANAS a total of three times during the second online session. The first time was at the beginning of the session to assess participants’ baseline state affect. The second was after participants recalled an academic failure. These first two administrations formed a manipulation check to assess whether people actually became distressed by the recollection of their failures, as evidenced by changes in their PA and NA from pre- to postrecall. Finally, we administered the PANAS a third time immediately after the participants listened to the audio clips to examine the impact of hearing the different audio clips on PA or NA. In the current sample, at baseline, Cronbach’s 162 MILLER AND KELLY alpha for PA and NA were both .91 and the positive and negative scales were uncorrelated, r ⫽ .04, p ⫽ .69. State self-compassion. State self-compassion was assessed in two ways. First, coded written behaviour measured postmanipulation levels of state self-compassion, which we modelled after the methods employed by Sbarra, Smith, and Mehl (2012). Two trained coders independently assessed participants’ written account of their reaction to a recent negative academic event. Coders rated each description according to the 12-item Self-Compassion Scale-Short Form (SCS-SF; Raes, Pommier, Neff, & Van Gucht, 2011) in order to be consistent with the protocol used in previous research coding qualitative self-compassion data (Sbarra et al., 2012). Wording of the items in this scale were modified to assess someone else’s self-compassion. Sample coding items were: “In describing their failure this person tried to take a balanced view of the situation.” Coders were undergraduate students who were familiar with the construct of self-compassion and its facets. Kathryn Miller led two training sessions for the coders during which they practiced using the coding protocol. She provided coders with a written description of the six components of self-compassion as well as example statements that might be rated highly for a given component for reference. Coders were instructed to provide ratings based on what proportion of sentiments in the paragraph expressed the component and how strongly the sentiment was expressed using a 5-point Likert scale from 1 (Strongly Disagree) to 5 (Strongly Agree). If one of the items (i.e., trying to take a balanced view of the situation) was absent from the description, coders were instructed to rate the item a 3 (Neither Agree nor Disagree). Item scores were averaged with negative items reverse-scored. Each participant therefore received two scores, one per coder; the mean of these two scores yielded one overall score indicating how selfcompassionately they described their failure. A two-way random intraclass correlation assessed the agreement between coders. According to commonly cited guidelines outlined by Cicchetti (1994) and Hallgren (2012), which classify an ICC of between .75 and 1.00 as “excellent”, a high degree of reliability was found between the two coders, ICC (2, 2) ⫽ .79, 95% CI [.70, .85] indicating that 79% of the variance in the mean of the ratings was due to shared variance between the two coders in their assessment of written self-compassion. Following the writing task, participants’ self-reported state selfcompassion was measured using an adapted version of the SCS modified by Breines and Chen (2013) in order to be consistent with previous research assessing state self-compassion in response to a specific event/struggle. Breines and Chen modified the original SCS to assess state self-compassion regarding a specific failure, and shortened the scale to 16 items, as many of the original items designed to measure trait self-compassion did not apply to recalling a specific failure. The modified scale used in the present study instructed participants to respond to these 16 items based on their current reaction to a recalled personal failure that is, “I am trying to take a balanced view of the situation” and “I am trying to be kind and reassuring to myself” on a Likert scale from 1 (Strongly Disagree) to 7 (Strongly Agree). The mean of all items, with negative items reverse-scored, formed our measure of selfreported state self-compassion. The Cronbach’s alpha was .91. State self-esteem. State self-esteem was measured postmanipulation by administering a state version of the RSE that in- structed participants to respond to RSE items according to their feelings in the current moment, rather than how they generally feel about themselves. The Cronbach’s alpha was .93. As this study investigates the unconscious, implicit effects of hearing a display of self-compassion, premanipulation measures of state self-compassion and self-esteem were not included in order to avoid priming a self-compassionate or self-esteem focused orientation in the participants before even hearing the audio clips. Engagement and attention measures. Two single-item measures were included at the end of the second session to assess participants’ engagement and attention. Participants rated how engaged they were in the study tasks from 0 (Not Engaged at All) to 100 (Totally Engaged) and how much attention they paid to the audio clip from 1 (Did Not Pay Any Attention) to 4 (Paid Very Close Attention). Procedure In order to test our hypothesis that self-compassion would be contagious, we felt it would be important for participants to be unaware of the main study objective. As such, partial disclosure was required. The description posted on the online research participation pool and the information consent letter stated that this study was investigating “how university students cope with academic failure.” Participants completed two online sessions 1 week apart. Participants accessed the first session through a link to a Qualtrics survey via the online participant pool and completed this online session by filling out trait measures of self-compassion and selfesteem. At the end of that session, we informed participants that in 1 week they would be e-mailed the link to access the second session. They had 48 hr in which to complete this portion of the study once they had received the link. The second online session started by measuring baseline affect. We then instructed participants to bring to mind a recent academic failure or negative academic experience. In order to encourage participants to thoroughly visualize their failure, they had 1 minute to recall this memory, and were unable to move on to the next page of the study until the full minute had elapsed. After the minute of recall, participants reported on their affect again. Next was the experimental manipulation in which participants were randomly assigned to listen to an audio clip of a female voice describing a personal academic failure she had recently experienced in a selfcompassionate way, a self-esteem enhancing way, or a factual and objective way. Cover story. Before hearing the clip, we informed participants that they would soon be writing about their own academic failure and that the purpose of the clip was to demonstrate the amount of detail to include in their own description. We included this cover story to encourage the participant to listen in a general way to the speaker to simulate how one might listen to social interaction partners in daily life. We opted not to encourage a specific listening style focused on the actor’s coping style so we could better examine whether self-compassion might be contagious at an implicit level and so we could prevent participants from guessing, and behaving according to, study hypotheses. Experimental manipulation. The three audio clips followed the same structure, and were approximately 2.5 min each in length. Each clip started with a report of how the individual studied hard IS SELF-COMPASSION CONTAGIOUS? for an upcoming test, felt well prepared, and then at the test, completely blanked, panicked, and barely passed. The clips then described how the individual felt afterward. The clips were designed by the researchers and were intended to be face valid and theoretically consistent with literature on the components of selfcompassion and self-esteem enhancement, respectively, with the control condition being an emotionally neutral description of the situation. We further ensured that we depicted the positive cognitive and affective outcomes that have been empirically linked with each respective construct, as described below. We revised and edited the clips numerous times based on feedback from other researchers with expertise in self-compassion and self-esteem, and based on ratings of the audio clips provided by eight students with no familiarity with this area of research. Self-compassion condition. Participants who were assigned to the self-compassion condition heard someone describe a selfcompassionate reaction including Neff’s (2003) three components of self-kindness, mindfulness, and common humanity. The actor spoke in a warm and soothing tone of voice, characteristic of compassion (Gilbert, 2005), and described a sense of connection with others, and feelings of reassurance and acceptance about the situation. Below is an excerpt from this clip: I said to myself, “You know what? I see that you are hurting and I am sorry this experience was upsetting for you (self-kindness). I understand that you are disappointed—that’s so natural after an experience like this. Try to give yourself permission to feel upset. . . . These are feelings that will pass with time (mindfulness), and that most people would experience after a setback (common humanity).” . . . Looking at the larger picture I was able to feel much more reassured. I recognised that I won’t feet discouraged forever and was able to feel more accepting of the situation. Self-esteem enhancement. Participants in the self-esteem enhancing condition heard a reaction that was intended to help boost the actor’s own self-worth using an energizing tone, and reflecting components known to be associated with self-esteem enhancement such as positive self-beliefs, downward social comparisons, and the tendency to make external attributions for failures and evaluate the reasons for failure in a self-serving way. These elements function to preserve positive self-evaluation and disengage from the negative affect associated with failures (for review see Crocker & Park, 2004). An excerpt from this clip is: I realised there was no cause for disappointment given that I still managed to score as good or better than over half the class (downward social comparison). Everyone else found it hard too, so clearly I deserve to be here at university just as much or more than any other student in that class (evaluating evidence in self-serving way). I know I’m still a worthwhile person (positive self-views). I have lots of good qualities—I am smart, hard-working, and doing well in all my other courses . . . and I still feel good about myself despite this one test. I told myself “it just goes to show, you can study all the time in the world, but if it’s an unfair test there’s not much you can do about it (external attribution). You are still a smart, popular leader amongst your peers, shake it off (positive self-view)!” Control condition. Participants in the control condition heard a reaction that was factual and objective, without any reflective component, and with a neutral tone of voice. An excerpt from this clip is below: 163 The essay questions were a little better because I could write whatever I wanted, but I was so worked up at this point I wasn’t even sure if what I was writing made any sense. At the end of the exam the prof collected our papers and I left as quickly as I could. I ended up barely scraping by, passing, but not by much. After listening to their assigned audio clip, participants selfreported on their current affect again. They were then instructed to bring back to mind the negative academic event they had previously recalled, and to provide a written description of the event including details of what happened, their reaction at the time, and how they feel about it currently upon reflection. Participants then completed measures of state self-compassion, state self-esteem, and engagement and attention. Finally, they received a debriefing form explaining the true purpose of the study, which condition they were in, and why partial disclosure was necessary. Results We conducted all analyses in IBM SPSS Statistics 22 using mixed factorial analyses of variance (ANOVAs) or analyses of covariance (ANCOVAs). The main independent variable was condition (hearing a self-compassionate, self-esteem enhancing, or control audio clip). The main dependent variables were state self-compassion, state self-esteem, PA, and NA. In all analyses examining state levels of self-compassion and self-esteem as the dependent variables, trait self-compassion and trait self-esteem were included as initial covariates. Based on convention in past empirical research (Allen & Leary, 2014; Leary et al., 2007) and the correlations obtained in the current study (see below), we expected that each of these variables would contribute unique variance to participants’ state levels of self-compassion and selfesteem. When trait self-compassion and/or trait self-esteem did not contribute significantly to one of these dependent variables, we removed the relevant covariate from the final models. When examining the effect of condition on PA and NA, premanipulation levels of the relevant affect variable served as a covariate with postmanipulation levels serving as the dependent variable. When assessing skewness, kurtosis, and outliers in the sample, all variables appeared to be normally distributed. To examine the intercorrelations among state and trait self-compassion and selfesteem, we conducted Pearson zero-order correlations. Trait selfesteem and trait self-compassion were strongly correlated, r ⫽ .74, p ⬍ .001, which is consistent with previous research. Postmanipulation state self-compassion correlated with trait self-compassion, r ⫽ .55, p ⬍ .001 and state self-esteem correlated with trait self-esteem, r ⫽ .79, p ⬍ .001. Additionally trait self-compassion correlated with state self-esteem (r ⫽ .60, p ⬍ .001) and trait self-esteem correlated with state self-compassion, r ⫽ .60, p ⬍ .001. These correlations supported our decision to control for trait self-compassion and trait self-esteem in initial ANCOVAs examining state self-compassion and state self-esteem as dependent variables. Preliminary Analyses Baseline differences between conditions. Table 1 presents means and standard deviations for all variables at baseline/premanipulation by condition. As seen in Table 1, one-way ANOVAs revealed that there were no differences in levels of trait self- MILLER AND KELLY 164 Table 1 Means (and Standard Deviations) of Study Variables at Baseline Variable a Trait self-compassion Trait self-esteemb Premanipulation NAc Premanipulation PAd a n ⫽ 106. b n ⫽ 111. c Self-compassion Self-esteem Control Effect of condition 2.33 (.10) 32.19 (1.37) 20.79 (1.66) 17.60 (1.26) 2.45 (.10) 35.73 (1.37) 22.94 (1.70) 20.63 (1.29) 2.52 (.10) 32.19 (1.37) 21.92 (1.66) 20.16 (1.24) F(2, 103) ⫽ 1.00, p ⫽ .37 F(2, 108) ⫽ 1.78, p ⫽ .17 F(2, 109) ⫽ .41, p ⫽ .66 F(2, 107) ⫽ 1.67, p ⫽ .19 n ⫽ 112. d n ⫽ 110. compassion, trait self-esteem, premanipulation NA, or premanipulation PA across conditions. Pre- and postrecall affect. The experimental manipulation was contingent on participants recalling a distressing personal academic experience, and we instructed them to recall the experience in as vivid detail as possible. In order to verify that the recall portion of the manipulation worked as intended, we conducted repeated measures ANOVAs to examine changes in PA and NA from pre- to postrecall. Analyses revealed that the manipulation appeared to sufficiently engage the participants’ emotions. There was a significant effect of time on NA and PA. NA increased significantly across conditions from pre- to postrecall, F(1, 106) ⫽ 32.20, p ⬍ .001, p2 ⫽ .233 and PA decreased significantly across conditions from pre to post recall (F(1, 105) ⫽ 32.455, p ⬍ .001, p2 ⫽ .236). There was no condition by time interaction effect in predicting either NA, F(2, 106) ⫽ .387, p ⫽ .680, p2 ⫽ .007 or PA (F(2, 105) ⫽ 1.407, p ⫽ .249, p2 ⫽ .026). Engagement and attention. There were 12 outliers with ratings ⬎3 SD below the mean for either engagement or attention; however, the pattern of results did not change when they were excluded from analyses, thus we chose to retain them in the final analyses. The average rating of engagement in the study was 70 out of 100 (SD ⫽ 22.88) and there were no differences in participant engagement across conditions, F(2, 114) ⫽ 1.19, p ⫽ .31. The mean rating of attention paid during the audio clip was 3.4 out of 4 (SD ⫽ 0.72), and there were no differences in attention paid across conditions F(2, 115) ⫽ 0.91, p ⫽ .41. Therefore, participants were reasonably engaged and motivated to complete the study tasks and attended closely to their assigned audio clip. Main Analyses Effect of condition on state self-compassion. In order to assess whether hearing an actor describe a failure selfcompassionately would yield higher state self-compassion than hearing a self-esteem enhancing or neutral description, we conducted two factorial ANCOVAs. In the first, the dependent variable was observer-rated self-compassion levels based on participants’ written accounts of their failure, and in the second, self-reported self-compassion served as the dependent variable. Observer-coded written state self-compassion. Trait selfcompassion was not a significant predictor in the initial model, F(1, 95) ⫽ 0.76, p ⫽ .40 and so was removed. In the final model, when controlling for trait self-esteem, F(1, 106) ⫽ 9.75, p ⫽ .002, p2 ⫽ .08, there was a medium-sized main effect of condition on observer-rated self-compassion, F(2, 106) ⫽ 3.97, p ⫽ .022, p2 ⫽ .07. As shown in Figure 1, and in partial support of Hypothesis 1a, contrasts revealed that although those in the self-compassion and Figure 1. Participants’ mean estimated written self-compassion as a function of experimental condition. Objective ratings of written self-compassion were higher in the self-compassion condition than the control condition, but did not differ between the self-esteem and control conditions. IS SELF-COMPASSION CONTAGIOUS? self-esteem conditions did not differ significantly in how selfcompassionately they wrote about their personal failure and the effect size was small (p ⫽ .37, p2 ⫽ .007), those in the self-compassion condition wrote significantly more self-compassionately than those in the control condition (p ⫽ .007, p2 ⫽ .07) with a medium effect, whereas those in the self-esteem condition did not, and the effect size for this contrast was small (p ⫽ .067, p2 ⫽ .03). See Table 2 for mean differences between conditions. Self-reported state self-compassion. Controlling for trait selfcompassion and trait self-esteem, each of which contributed significantly (F ⫽ 0.57, p ⫽ .02, p2 ⫽ .063 and F ⫽ 9.47, p ⫽ .003, p2 ⫽ .10), there was a medium main effect of condition on self-reported state self-compassion, F(2, 85) ⫽ 4.69, p ⫽ .01, p2 ⫽ .10. As shown in Figure 2, contrasts partially supported Hypothesis 1b. Participants in the self-compassion and self-esteem conditions did not differ significantly in postmanipulation levels of state self-compassion (p ⫽ .20, p2 ⫽ .02) and the effect size was small; however, participants in the self-compassion condition reported significantly higher state self-compassion than those in the control condition (p ⫽ .003, p2 ⫽ .10) with a medium effect. Conversely, those in the self-esteem condition did not differ significantly in their level of self-reported state self-compassion from those in the control condition and the effect size for this contrast was small (p ⫽ .067, p2 ⫽ .04). See Table 2 for mean differences between conditions. Effect of condition on negative affect. An ANCOVA revealed support for Hypothesis 2 that hearing a self-compassionate account of failure would yield lower NA than hearing one of the other descriptions. Controlling for preclip NA, F(1, 105) ⫽ 69.80, p ⬍ .001, p2 ⫽ .40, there was a significant medium-sized effect of condition on postclip NA, F(2, 105) ⫽ 4.54, p ⫽ .013, p2 ⫽ .08. Table 2 Multiple Comparisons and Mean Differences in Dependent Variables Comparison Mean difference SE 95% CI a Observer-coded state self-compassion Self-compassion vs control 4.74ⴱ 1.72 Self-compassion vs self-esteem 1.56 1.74 Self-esteem vs control 3.1 1.72 Self-reported state self-compassionb Self-compassion vs. control 10.86ⴱ 3.58 Self-compassion vs. self-esteem 4.55 3.52 Self-esteem vs. control 6.31 3.40 Negative affectc Self-compassion vs. control ⫺4.00ⴱ 1.33 Self-compassion vs. self-esteem ⫺2.28 1.35 Self-esteem vs. control ⫺1.71 1.35 Positive affectd Self-compassion vs. control 3.49ⴱ 1.15 Self-compassion vs. self-esteem ⫺2.02 1.17 ⴱ Self-esteem vs. control 5.50 1.15 a 1.33, 8.15 ⫺1.90, 5.01 ⫺.23, 6.59 3.73, 18.00 ⫺2.46, 11.56 ⫺.45, 13.06 ⫺6.62, ⫺1.36 ⫺4.96, .40 ⫺4.39, .96 1.21, 5.76 ⫺4.34, .31 3.22, 7.79 Comparisons based upon ANCOVA adjusted means controlling for Trait Self-Esteem mean of 34.22. b Comparisons based on ANCOVA adjusted means controlling for Trait Self-Esteem mean of 33.97 and Trait SelfCompassion mean of 2.44. c Comparisons based on ANCOVA adjusted means controlling for preclip Negative Affect mean of 21.86. d Comparisons based on ANCOVA adjusted means controlling for preclip Positive Affect mean of 19.52. ⴱ p ⬍ .05. 165 As shown in Figure 3, contrasts revealed that postclip NA in the self-compassion condition did not differ significantly from the self-esteem condition (p ⫽ .09, p2 ⫽ .03) with small effect. However, the self-compassion condition yielded significantly lower postclip NA than the control condition (p ⫽ .003, p2 ⫽ .08) with a medium effect, but the self-esteem did not (p ⫽ .21, p2 ⫽ .02) with a small effect. See Table 2 for mean differences between conditions. Effect of condition on positive affect. Results supported Hypothesis 3 that both the self-esteem and self-compassion conditions would yield higher PA than the control condition. As shown in Figure 4, an ANCOVA revealed that when controlling for preclip PA, F(1, 101) ⫽ 175.10, p ⬍ .001, p2 ⫽ .63, there was a large main effect of condition on postclip PA, F(2, 101) ⫽ 11.79, p ⬍ .001, p2 ⫽ .19. Contrasts showed that the self-compassion condition and the self-esteem condition each produced significantly higher postclip PA than the control condition (p ⫽ .003, p2 ⫽ .08 and p ⬍ .001, p2 ⫽ .19, respectively). There were no significant differences between the self-esteem and self-compassion conditions (p ⫽ .09, p2 ⫽ .03). See Table 2 for mean differences. Exploratory Analysis of Condition Effect on State Self-Esteem An additional exploratory analysis was conducted to determine if self-esteem might also be “contagious”—that is, if participants in the self-esteem condition would experience greater postclip state self-esteem than those in the self-compassion and control conditions. An ANCOVA revealed that trait self-compassion did not significantly contribute to variance in the model, F(1, 95) ⫽ 0.58, p ⫽ .45, p2 ⫽ .006. In the final model that only controlled for trait self-esteem, F(1, 106) ⫽ 175. 32, p ⬍ .001, p2 ⫽ .62, there was no effect of condition on state self-esteem, F(2, 106) ⫽ 0.67, p ⫽ .51, p2 ⫽ .01. Therefore, state levels of self-esteem did not differ based on the audio clip participants heard. Discussion The current study was the first to test the novel idea that self-compassion might be contagious. Specifically, we investigated the differing impact of hearing a peer respond to an academic failure in a self-compassionate, self-esteem enhancing, or factual manner. Results showed partial support for our primary hypothesis. Consistent with Hypothesis 1, compared to participants in the control condition, participants who heard someone display self-compassion subsequently reported experiencing more self-compassion when recalling their own personal failure and also wrote about their failure in an objectively more self-compassionate way. Although those in the self-esteem condition did not differ from the control condition in these ways, there were no significant differences between the self-esteem and self-compassion conditions in self-reported or objective self-compassion, contrary to Hypothesis 1. Our second hypothesis also evidenced partial support, as participants in the self-compassion condition reported lower NA than those in the control condition whereas those in the self-esteem condition did not, although again, there were no significant differences between the self-esteem and self-compassion conditions. Our third hypothesis was supported, as participants who heard either a self-esteem enhancing or self-compassionate account of 166 MILLER AND KELLY Figure 2. Participants’ mean estimated state self-compassion as a function of experimental condition. State self-compassion was higher in the self-compassion condition than the control condition, but did not differ between the self-esteem and control conditions. failure had subsequently higher PA than those who heard a neutral account. Finally, postmanipulation state self-esteem did not differ across conditions suggesting that self-compassion was transmissible across individuals but state self-esteem was not. These findings are the first to suggest that self-compassionate reports of failure by one person may yield higher levels of self-compassion and improved affect in the listener as compared to neutral descriptions of a personal failure. Although these results support our overall hypothesis that self-compassion and its associated emotional benefits may be contagious, there were no significant differences between the self-compassion and self-esteem conditions in postclip levels of self-compassion and NA. It is unclear why this may have been. One possibility is that the audio clips in these two conditions differed, but not sufficiently to produce large enough differences in participants’ reactions. This may be in part due to the fact that the constructs of self-compassion and self-esteem are positively correlated with one another and thus although the clips had important differences, they had certain elements in common (e.g., positive reappraisal, emotion regulation to reduce distress). It may be that a within-persons design in which participants hear both a selfcompassionate and self-esteem enhancing display back-to-back Figure 3. Mean estimated postclip negative affect, controlling for preclip levels, as a function of experimental condition. Postclip negative affect was lower in the self-compassion condition than the control condition, but did not differ between the self-esteem and control conditions. IS SELF-COMPASSION CONTAGIOUS? 167 Figure 4. Mean estimated postclip positive affect, controlling for preclip levels as a function of experimental condition. Participants in the self-compassion and self-esteem conditions had significantly higher positive affect than those in the control condition. would yield more distinctive reactions to each clip as compared to the differences observed between conditions in the present between-persons design. A second possibility may be that reactions to hearing a selfesteem enhancing versus self-compassionate coping style would differ in a more pronounced way when a self-esteem enhancing display employs more problematic methods of boosting self-worth than those used in our clip such as narcissism, aggressiveness, and extreme defensiveness (Crocker & Park, 2004). Finally, it is possible that our study cover story minimised the differences between the reactions elicited by the self-compassion and self-esteem conditions. That is, we instructed participants to listen to the clips in order to get a sense of how much detail to include in their personal failure descriptions but did not explicitly ask them to attend to the specific coping style being displayed. In future research, it will be important to examine whether encouraging more careful listening to the actor’s coping style produces differences between those who listen to self-esteem-enhancing versus self-compassionate accounts of failure. It is worth noting that both the self-esteem and self-compassion conditions yielded medium effect sizes in terms of their impact on PA relative to the control condition. However, for state selfcompassion and NA the effect of hearing a self-compassionate description of failure compared to a neutral one was medium and significant, whereas the difference between hearing a self-esteem enhancing description compared to a neutral one was small and nonsignificant. Taken together these results suggest two things. First, even though there may have been overlap between the self-esteem and self-compassion conditions (i.e., positive coping), they were distinct enough from one another to produce different patterns of results when compared to the control condition for both state self-compassion, which was our primary outcome of interest, as well as for NA. Second, it seems that any display of positive self-regard in the face of failure by others may benefit one’s own positive affect, but one’s own self-compassion and NA may be especially responsive to displays of self-compassion. Theoretical Explanations How did hearing a self-compassionate account of failure positively affect participants’ self-compassion and NA more than hearing a neutral account of failure? One possibility is that participants consciously or unconsciously detected the particular self-attitude being portrayed while listening to the clip and that this awareness influenced their subsequent feelings and behaviour either in a conscious or implicit fashion. Both of these interpretations would be consistent with Social Learning Theory (Bandura, 1971) and simulation theory. That is, listeners may have noticed that the person who displayed self-compassion also exhibited lower distress. This observed affective benefit may have served as vicarious reinforcement and motivation to employ self-compassion in one’s own self-relating, and/or may have generated an internal simulation of self-compassion via mirror neurons. Our findings additionally suggest that listeners may also experience lower NA themselves from simply hearing a self-compassionate display. Thus, in addition to self-compassion itself, the benefits associated with being selfcompassionate may be contagious. Future research should explore this idea, and should examine whether consistent with Social Learning Theory, self-compassion contagion happens only when witnessing positive consequences of self-compassion (i.e., someone feeling better after treating themselves self-compassionately) and not when observing negative consequences (i.e., someone being belittled for displaying self-compassion). 168 MILLER AND KELLY It is interesting to note that hearing a display of self-esteem did not lead to higher self-esteem in the listener than the other two conditions. Our pattern of findings, which implies that selfcompassion may be “contagious” but self-esteem is not, suggests that observers may not assimilate all types of intrapersonal processes they observe. Perhaps only intrapersonal variables that promote consideration of others are amenable to the contagion effect. For example, it may be that features of the self-compassionate display that emphasised common humanity—the recognition that others struggle too— helped listeners to detect and assimilate the self-compassionate mindset of the actor. Self-esteem has no such dimension, and indeed maintaining high self-esteem can sometimes necessitate distancing oneself from, or derogating others (Crocker, Thompson, McGraw, & Ingerman, 1987; Tesser, 2000). We nevertheless found that hearing either a self-esteem enhancing or selfcompassionate account of failure was more beneficial for the listener’s PA than hearing a neutral account, suggesting that hearing displays of positive self-attitudes in general may boost general positive feelings in the listener. It will be important for future research to replicate our findings and to determine which intrapersonal variables, and associated “rewards,” are most contagious and why. Practical Implications Although replication is needed, our findings suggest that in a student population, exposure to displays of self-compassion from others may be a preliminary way to raise self-compassion levels. Given the medium-to-large effect sizes found in the current study, an important next step will be to employ a naturalistic observational design to examine how the magnitude of these effects may change in a real-life interaction, and to explore whether and how the benefits of hearing self-compassion are subjectively felt and utilized. This line of research would suggest that encouraging peers and individuals responsible for social programming (i.e., teachers, counselors and social workers) to model self-compassion may help increase self-compassion and improve affect when the resources and expertise to implement formal self-compassion interventions are unavailable. Similarly, it may be useful to examine self-compassion modelling in the family context. Research shows that self-criticism in parents relates to self-criticism in children (Amitay, Mongrain, & Fazaa, 2008; Thompson & Zuroff, 1998), so displaying selfcompassion may be an important way for parents to combat self-criticism in their child. Future research should further examine the contexts within which self-criticism and self-compassion may be contagious. Such findings may also extend to teachers, guidance counselors, and early childhood educators. Our results additionally suggest that modelling self-esteem may also yield emotional benefits, namely heightened PA, but this modelling is unlikely to influence others’ self-esteem. From a clinical standpoint, our findings suggests that although explicitly teaching selfcritical clients to increase their self-compassion through formal interventions is the most empirically supported way to raise selfcompassion (Gilbert & Irons, 2004; Gilbert & Procter, 2006; Kelly & Carter, 2015; Kelly, Zuroff, & Shapira, 2009), encouraging these clients to surround themselves with self-compassionate others may be a useful addition to formal self-compassion training. Indeed, self-critical individuals may be resistant to cultivating self-compassion in a more explicit manner for fear that it will cause them to lower their standards or become complacent in some way (Gilbert, McEwan, Matos, & Rivis, 2011). Future research should investigate whether for these individuals, being in a selfcompassionate interpersonal context may be an alternative way for them to experience an increase in their own self-compassion. This study is one of the first to show that current interpersonal contexts may affect self-compassion levels. Breines and Chen (2013) found that personal self-compassion levels increased in response to providing care to others. Our study suggests that another way to raise self-compassion may be to surround oneself with people who treat themselves positively in the face of personal failure or distress. Such information is valuable in light of recent research suggesting that the act of extending care to others may not always be associated with increased self-compassion and in some cases may undermine it (Hermanto & Zuroff, 2016). Limitations and Future Directions There were a number of limitations in this study that should be addressed in future research. First, this was a very brief experimental manipulation. Employing a more naturalistic observational study design will be important to test the duration and ecological validity of the self-compassion contagion effect. It will also be important to determine how lasting the self-compassion contagion effect is, and whether individuals can better sustain higher levels of self-compassion when surrounded by self-compassionate others on a more continual basis. Second, this was a homogeneous sample of undergraduate females with a modest sample size (N ⫽ 119). Research has shown gender and age differences in self-compassion (Neff & Vonk, 2009; Yarnell et al., 2015); thus, it will be important to test the self-compassion contagion effect in larger, more diverse samples to increase power and ensure generalizability. Testing the contagion effect in a clinical sample of people with pathological selfcriticism will also be important in the current line of research. It would also be important to examine whether demographic variables such as ethnicity, age, and gender influence these effects. The current study also focused on academic failure when inducing self-compassion, and thus inducing the self-compassion contagion effect in other domains will be important to establish the scope of the effect. It could be that for an especially distressing failure, observing a self-compassionate display in another would not be enough to raise one’s own self-compassion. Evidently there is rich material for future research on the contextual variables that may moderate the self-compassion contagion effect. Conclusions This was the first study to find that self-compassion may be contagious. On a theoretical level, this study adds to the empirical literature in social psychology suggesting that intrapersonal experiences can be transmitted interpersonally (Bandura & Whalen, 1966; de Vignemont & Singer, 2006; Oh & Damhorst, 2009), and may affect levels of that same variable in others. Indeed, the results of this study suggest that the “contagious” adoption of selfcompassion by the listener may manifest both explicitly, through her self-report, and implicitly, through her behaviour. Our findings also add to the limited literature that highlights the influence of IS SELF-COMPASSION CONTAGIOUS? one’s current interpersonal context on one’s levels of selfcompassion, and suggest that there may be merit in exploring other contextual factors that might have a similar impact. It will also be important in future research to identify how consciously versus unconsciously the transfer of self-compassion occurs, what factors moderate and drive the effect, and whether the mechanisms differ in brief interactions versus extended social relationships. Our preliminary results nevertheless suggest that modelling selfcompassionate behaviour may be an effective way to encourage self-compassion in others, and that surrounding oneself with selfcompassionate others may be an effective way to increase one’s own level of self-compassion. Résumé L’autocompassion étant associée à de multiples émotions et avantages sur le plan physique, il importe d’évaluer les contextes sociaux qui ont une influence sur celle-ci. Cette étude était la première à examiner si l’exposition à l’autocompassion manifestée par autrui avait un effet sur le niveau d’autocompassion de la personne exposée. On a ainsi demandé à cent dix-neuf étudiantes universitaires d’entendre diverses réactions, soit d’autocompassion, de renforcement de l’estime de soi et neutres, face à un échec scolaire, d’écrire à propos de leur propre échec, puis de mesurer leur propre niveau d’autocompassion et d’affect. Les résultats ont démontré que le fait d’entendre une réaction d’autocompassion après s’être remémoré un échec scolaire personnel exacerbait le sentiment d’autocompassion, selon un questionnaire d’auto-évaluation et des réponses écrites codées, et abaissait l’affect négatif en comparaison de l’écoute d’une réaction factuelle, alors que le fait d’entendre une réaction de renforcement de l’estime de soi n’avait pas cet effet. Le fait d’entendre une réaction d’autocompassion n’aboutissait pas à des résultats très différents en comparaison de la réaction au renforcement de l’estime de soi. Ces constatations sont les premières qui laissent entendre que le comportement d’autocompassion observé chez autrui peut influer sur la compassion qu’une personne peut ressentir envers elle-même. Mots-clés : autocompassion, estime de soi, interactions sociales, contagion affective, adaptation. References Ackerman, J. M., Goldstein, N. J., Shapiro, J. R., & Bargh, J. A. (2009). You wear me out: The vicarious depletion of self-control. Psychological Science, 20, 326 –332. http://dx.doi.org/10.1111/j.1467-9280.2009 .02290.x Adams, C. E., & Leary, M. R. (2007). Promoting self-compassionate attitudes toward eating among restrictive and guilty eaters. Journal of Social and Clinical Psychology, 26, 1120 –1144. http://dx.doi.org/10 .1521/jscp.2007.26.10.1120 Allen, A. B., & Leary, M. R. (2014). Self-compassionate responses to aging. The Gerontologist, 54, 190 –200. http://dx.doi.org/10.1093/ geront/gns204 Amitay, O. A., Mongrain, M., & Fazaa, N. (2008). Love and control: Self-criticism in parents and daughters and perceptions of relationship partners. Personality and Individual Differences, 44, 75– 85. http://dx .doi.org/10.1016/j.paid.2007.07.020 Badaly, D. (2013). Peer similarity and influence for weight-related outcomes in adolescence: A meta-analytic review. Clinical Psychology Review, 33, 1218 –1236. http://dx.doi.org/10.1016/j.cpr.2013.09.003 169 Bandura, A. (1971). Social learning theory. New York, NY: General Learning Press. Bandura, A., & Whalen, C. K. (1966). The influence of antecedent reinforcement and divergent modeling cues on patterns of self-reward. Journal of Personality and Social Psychology, 3, 373–382. http://dx.doi .org/10.1037/h0023022 Barnard, L., & Curry, J. (2011). Self-compassion: Conceptualizations, correlates, and interventions. Review of General Psychology, 15, 289 – 303. http://dx.doi.org/10.1037/a0025754 Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, 103, 5–33. http://dx.doi.org/10.1037/0033-295X .103.1.5 Breines, J. G., & Chen, S. (2013). Activating the inner caregiver: The role of support-giving schemas in increasing state self-compassion. Journal of Experimental Social Psychology, 49, 58 – 64. http://dx.doi.org/10 .1016/j.jesp.2012.07.015 Breines, J., Toole, A., Tu, C., & Chen, S. (2014). Self-compassion, body image, and self-reported disordered eating. Self and Identity, 13, 432– 448. http://dx.doi.org/10.1080/15298868.2013.838992 Cicchetti, D. V. (1994). Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment, 6, 284 –290. http://dx.doi.org/10.1037/10403590.6.4.284 Crocker, J., & Park, L. E. (2004). The costly pursuit of self-esteem. Psychological Bulletin, 130, 392– 414. http://dx.doi.org/10.1037/00332909.130.3.392 Crocker, J., Thompson, L. L., McGraw, K. M., & Ingerman, C. (1987). Downward comparison, prejudice, and evaluations of others: Effects of self-esteem and threat. Journal of Personality and Social Psychology, 52, 907–916. http://dx.doi.org/10.1037/0022-3514.52.5.907 Crocker, J., & Wolfe, C. T. (2001). Contingencies of self-worth. Psychological Review, 108, 593– 623. http://dx.doi.org/10.1037/0033-295X .108.3.593 de Vignemont, F., & Singer, T. (2006). The empathic brain: How, when and why? Trends in Cognitive Sciences, 10, 435– 441. http://dx.doi.org/ 10.1016/j.tics.2006.08.008 Gallese, V., Eagle, M. N., & Migone, P. (2007). Intentional attunement: Mirror neurons and the neural underpinnings of interpersonal relations. Journal of the American Psychoanalytic Association, 55, 131–175. http://dx.doi.org/10.1177/00030651070550010601 Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the premotor cortex. Brain, 119, 593– 609. http://dx.doi .org/10.1093/brain/119.2.593 Gilbert, P. (Ed.), (2005). Compassion: Conceptualisations, research, and use in psychotherapy. London, UK: Routledge. Gilbert, P., & Irons, C. (2004). A pilot exploration of the use of compassionate images in a group of self-critical people. Memory, 12, 507–516. http://dx.doi.org/10.1080/09658210444000115 Gilbert, P., McEwan, K., Matos, M., & Rivis, A. (2011). Fears of compassion: Development of three self-report measures. Psychology and Psychotherapy: Theory, Research and Practice, 84, 239 –255. http://dx .doi.org/10.1348/147608310X526511 Gilbert, P., & Procter, S. (2006). Compassionate mind training for people with high shame and self-criticism: Overview and pilot study of a group therapy approach. Clinical Psychology & Psychotherapy, 13, 353–379. http://dx.doi.org/10.1002/cpp.507 Goldstein, N. J., & Cialdini, R. B. (2007). The spyglass self: A model of vicarious self-perception. Journal of Personality and Social Psychology, 92, 402– 417. http://dx.doi.org/10.1037/0022-3514.92.3.402 Hallgren, K. A. (2012). Computing inter-rater reliability for observational data: An overview and tutorial. Tutorials in Quantitative Methods for Psychology, 8, 23–34. http://dx.doi.org/10.20982/tqmp.08.1.p023 170 MILLER AND KELLY Hefferline, R. F., & Keenan, B. (1963). Amplitude-induction gradient of a small-scale (covert) operant. Journal of the Experimental Analysis of Behavior, 6, 307–315. http://dx.doi.org/10.1901/jeab.1963.6-307 Hefferline, R. F., Keenan, B., & Harford, R. A. (1959). Escape and avoidance conditioning in human subjects without their observation of the response. Science, 130, 1338 –1339. http://dx.doi.org/10.1126/ science.130.3385.1338 Hermanto, N., & Zuroff, D. C. (2016). The social mentality theory of self-compassion and self-reassurance: The interactive effect of careseeking and caregiving. The Journal of Social Psychology, 156, 523– 535. http://dx.doi.org/10.1080/00224545.2015.1135779 Johnson, E. A., & O’Brien, K. A. (2013). Self-compassion soothes the savage ego-threat system: Effects on negative affect, shame, rumination, and depressive symptoms. Journal of Social and Clinical Psychology, 32, 939 –963. http://dx.doi.org/10.1521/jscp.2013.32.9.939 Kelly, A. C., & Carter, J. C. (2015). Self-compassion training for binge eating disorder: A pilot randomized controlled trial. Psychology and Psychotherapy: Theory, Research and Practice, 88, 285–303. http://dx .doi.org/10.1111/papt.12044 Kelly, A. C., & Stephen, E. (2016). A daily diary study of self-compassion, body image, and eating behavior in female college students. Body Image, 17, 152–160. http://dx.doi.org/10.1016/j.bodyim.2016.03.006 Kelly, A. C., Zuroff, D. C., & Shapira, L. B. (2009). Soothing oneself and resisting self-attacks: The treatment of two intrapersonal deficits in depression vulnerability. Cognitive Therapy and Research, 33, 301–313. http://dx.doi.org/10.1007/s10608-008-9202-1 Kennedy, T. D. (1970). Verbal conditioning without awareness: The use of programmed reinforcement and recurring assessment of awareness. Journal of Experimental Psychology, 84, 487– 494. http://dx.doi.org/10 .1037/h0029289 Kennedy, T. D. (1971). Reinforcement frequency, task characteristics, and interval of awareness assessment as factors in verbal conditioning without awareness. Journal of Experimental Psychology, 88, 103–112. http:// dx.doi.org/10.1037/h0030649 Kichler, J. C., & Crowther, J. H. (2009). Young girls’ eating attitudes and body image dissatisfaction. The Journal of Early Adolescence, 29, 212–232. http://dx.doi.org/10.1177/0272431608320121 Leary, M. R., Tate, E. B., Adams, C. E., Allen, A. B., & Hancock, J. (2007). Self-compassion and reactions to unpleasant self-relevant events: The implications of treating oneself kindly. Journal of Personality and Social Psychology, 92, 887–904. http://dx.doi.org/10.1037/ 0022-3514.92.5.887 Marston, A. R. (1965). Imitation, self-reinforcement, and reinforcement of another person. Journal of Personality and Social Psychology, 2, 255– 261. http://dx.doi.org/10.1037/h0022260 Neff, K. D. (2003). Development and validation of a scale to measure self-compassion. Self and Identity, 2, 223–250. http://dx.doi.org/10 .1080/15298860309027 Neff, K. D. (2016). The self-compassion scale is a valid and theoretically coherent measure of self-compassion. Mindfulness, 7, 264 –274. http:// dx.doi.org/10.1007/s12671-015-0479-3 Neff, K. D., & Beretvas, S. N. (2013). The role of self-compassion in romantic relationships. Self and Identity, 12, 78 –98. http://dx.doi.org/10 .1080/15298868.2011.639548 Neff, K. D., Kirkpatrick, K., & Rude, S. S. (2007). Self-compassion and adaptive psychological functioning. Journal of Research in Personality, 41, 139 –154. http://dx.doi.org/10.1016/j.jrp.2006.03.004 Neff, K. D., Tóth-Király, I., Yarnell, L., Arimitsu, K., Castilho, P., Ghorbani, N., . . . Mantios, M. (2019). Examining the factor structure of the Self-Compassion Scale in 20 diverse samples: Support for use of a total score and six subscale scores. Psychological Assessment, 31, 27– 45. http://dx.doi.org/10.1037/pas0000629 Neff, K. D., & Vonk, R. (2009). Self-compassion versus global selfesteem: Two different ways of relating to oneself. Journal of Personality, 77, 23–50. http://dx.doi.org/10.1111/j.1467-6494.2008.00537.x Neff, K. D., Whittaker, T., & Karl, A. (2017). Evaluating the factor structure of the Self-Compassion Scale in four distinct populations: Is the use of a total self-compassion score justified? Journal of Personality Assessment, 99, 596 – 607. Oh, K., & Damhorst, M. L. (2009). Coorientation of body image among older married couples. Body Image, 6, 43– 47. http://dx.doi.org/10.1016/ j.bodyim.2008.09.005 Raes, F., Pommier, E., Neff, K. D., & Van Gucht, D. (2011). Construction and factorial validation of a short form of the Self-Compassion Scale. Clinical Psychology & Psychotherapy, 18, 250 –255. http://dx.doi.org/ 10.1002/cpp.702 Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. http://dx.doi.org/10.1515/ 9781400876136 Sasmor, R. M. (1966). Operant conditioning of a small-scale muscle response. Journal of the Experimental Analysis of Behavior, 9, 69 – 85. http://dx.doi.org/10.1901/jeab.1966.9-69 Sbarra, D. A., Smith, H. L., & Mehl, M. R. (2012). When leaving your ex, love yourself: Observational ratings of self-compassion predict the course of emotional recovery following marital separation. Psychological Science, 23, 261–269. http://dx.doi.org/10.1177/0956797611429466 Tesser, A. (2000). On the confluence of self-esteem maintenance mechanisms. Personality and Social Psychology Review, 4, 290 –299. http:// dx.doi.org/10.1207/S15327957PSPR0404_1 Thompson, R., & Zuroff, D. C. (1998). Dependent and self-critical mothers’ responses to adolescent autonomy and competence. Personality and Individual Differences, 24, 311–324. http://dx.doi.org/10.1016/S01918869(97)00155-4 Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. http://dx.doi.org/10.1037/0022-3514.54.6.1063 Wicker, B., Keysers, C., Plailly, J., Royet, J. P., Gallese, V., & Rizzolatti, G. (2003). Both of us disgusted in my insula: The common neural basis of seeing and feeling disgust. Neuron, 40, 655– 664. http://dx.doi.org/ 10.1016/S0896-6273(03)00679-2 Yarnell, L. M., Stafford, R. E., Neff, K. D., Reilly, E. D., Knox, M. C., & Mullarkey, M. (2015). Meta-analysis of gender differences in selfcompassion. Self and Identity, 14, 499 –520. http://dx.doi.org/10.1080/ 15298868.2015.1029966 Zessin, U., Dickhäuser, O., & Garbade, S. (2015). The relationship between self-compassion and well-being: A meta-analysis. Applied Psychology: Health and Well-Being, 7, 340 –364. http://dx.doi.org/10.1111/ aphw.12051 Zuroff, D. C., Kelly, A. C., Leybman, M. J., Sadikaj, G., & Gilbert, P. (2012, January). Social safeness mediates the effects of self-compassion, self-criticism, and social support on interpersonal behaviour. Poster presented at the Annual Meeting of the Society for Personality and Social Psychology, San Diego, CA. Received January 5, 2018 Revision received August 10, 2019 Accepted August 12, 2019 䡲