Caregiving is a central component of human nature, representing a fundamental element of intimate relationships (Weiss, 1980). One particularly important type of caregiving is compassion. While definitions of compassion are varied, with different components being emphasized by different researchers and traditions (see Strauss et al., 2016 for an overview), compassion has often been defined as sensitivity to suffering in the self or others with a motivation to act to alleviate that suffering (Gilbert, 2014). The ability to give and receive compassion is fundamental to sustaining life (Gilbert, 2005; Goetz et al., 2010) and critical for thriving mental and physical health and wellbeing (Cosley et al., 2010; MacBeth and Gumley, 2012). According to Bowlby (1988), the capacity to develop meaningful emotional connections with others as a caregiver and care-seeker is a critical prerequisite for mental wellbeing. Importantly, a critical determinant of this capacity is believed to be a healthy attachment style (Bowlby, 1969; Gillath et al., 2005; Mikulincer and Shaver, 2017).

Attachment styles are emotional and behavioral patterns in close relationships with profound effects on psychological and social wellbeing (Bowlby, 1969; Hazan and Shaver, 1987) and internal working models of the self and others (see Bartholomew and Horowitz, 1991; Mikulincer, 1995). Differences in attachment in adults can be characterized in relation to two orthogonal dimensions: anxiety and avoidance, which are characterized by fear of rejection and discomfort with closeness, respectively. Those with low anxiety and avoidance are described as having a secure attachment characterized by comfort with closeness and confidence in consistent availability of support from others. In contrast, those high in either anxiety, avoidance, or both are described as insecure. Both anxious and avoidant insecure attachments are characterized by doubt in the availability of consistent emotional support from close others.

In the last 50 years, researchers have identified significant associations between attachments and the capacity to give compassion to others and oneself (for reviews, see Mikulincer and Shaver, 2003, 2007, 2017; Lathren et al., 2021; Varley et al., 2024). In relation to giving compassion to others, laboratory-based research shows that those with anxious attachments often provide compassion and caregiving less effectively than secure individuals, often engaging in compulsive caregiving – an overbearing and intrusive means of offering care and compassion (Kunce and Shaver, 1994; see Mikulincer and Shaver, 2017 for a review). This compulsive caregiving is believed to be driven by a tendency to be overly focused on one’s own personal distress and attachment needs for proximity and security (Collins and Read, 1994; Shaver and Mikulincer, 2002; Mikulincer and Shaver, 2017). Overall, most research suggests that anxious individuals are aware of needs for compassion in others and often endorse altruistic motivations for compassion (Feeney and Collins, 2003; Feeney et al., 2013), but provide compassion less effectively than secure individuals. However, there is also evidence suggesting that anxious individuals may be less responsive to needs for compassion in others. Specifically, Feeney and Collins (2001) identified that anxiously attached individuals were less likely to show support to a partner in need, especially when the partner’s support seeking is less obvious, and were less likely to provide effective emotional support. These findings show similarities to patterns observed in avoidantly attached individuals.

Studies suggest those with avoidant attachments provide compassion less often than securely attached individuals (Mikulincer and Shaver, 2017). Relative to more secure individuals, those high in avoidance report being less likely to maintain proximity and provide support to a partner in need and are often less sensitive to the needs of a partner (Simpson et al., 1992; Kunce and Shaver, 1994; Feeney and Collins, 2001; Millings et al., 2013; Péloquin et al., 2014). These tendencies are believed to be driven by negative beliefs about the availability and responsiveness of others, discomfort with closeness, and adaptive suppression of attachment needs and negative emotions (Shaver and Mikulincer, 2002; Mikulincer and Shaver, 2019). Overall, research suggests avoidant individuals provide compassion to others less often, and that anxious individuals provide compassion less effectively than those with secure attachments.

Insecurely attached individuals also report less self-compassion than secure individuals (see Lathren et al., 2021 for a review). Multiple studies have linked both anxiety and avoidance to lower self-compassion (e.g., Raque-Bogdan et al., 2011; Wei et al., 2011; Pepping et al., 2015; Joeng et al., 2017a,b; Homan, 2018; Brophy et al., 2020). Relatedly, those who experienced abuse in childhood or who recall early caregivers lacking warmth or showing signs of indifference or rejection also report lower self-compassion (Tanaka et al., 2011; Potter et al., 2014; Pepping et al., 2015; Westphal et al., 2016).

Links between attachment insecurity and difficulty receiving compassion from others have been less widely studied, but some research suggests a relationship. For example, research indicates those with insecure attachments often doubt the availability and supportiveness of others (Collins and Read, 1990; Collins, 1996; Kelly et al., 2012). A recent meta-analysis also identified that attachment insecurity is positively associated with self-reported fear and discomfort related to receiving compassion from others (Varley et al., 2024). Further, studies have identified that those with insecure attachments show physiological threat responses to imagery tasks involving imagining receiving compassion from others (Rockliff et al., 2008; Baldwin et al., 2020) and that those with low recalled parental warmth report discomfort receiving compassion (Kelly and Dupasquier, 2016). Together, these studies suggest a likely link between attachment insecurity and difficulty receiving compassion from others.

Overall, previous studies implicate attachment insecurity in difficulties with compassion. However, several questions remain unaddressed. First, while the deleterious effects of attachment insecurity on self-compassion and giving compassion to others have been more widely studied, experiences related to receiving compassion from others have received little attention. This means that while research indicates that insecurely attached individuals feel discomfort or threat when receiving compassion, it remains unknown if they are less aware of opportunities to receive compassion and less likely to allow others to act compassionately toward them. This is also problematic as receiving compassion from others is not necessarily highly correlated with self-compassion and giving compassion to others. Indeed, these types of compassion have been acknowledged as distinct constructs (e.g., Steindl et al., 2021). Accordingly, previously identified relationships between attachment and self-compassion or giving compassion may not necessarily correspond with relationships between attachment and receiving compassion.

It is critical we understand not just attachment-related differences in capacities for self-compassion and giving compassion to others, but also the capacity to receive compassion. This is because attachment insecurity’s negative impact on wellbeing may be partially driven by inability to derive soothing from others. Evidence illustrates that mental wellbeing and abilities to regulate emotions and respond to threat are partially reliant on the quantity and quality of interactions with others (Cacioppo and Patrick, 2008; Cozolino, 2014; Siegel, 2020). Bowlby (1988) also argued that the capacity to develop emotional connections with others as both a caregiver and care-seeker is critical for mental wellbeing. Given that insecurely attached individuals feel discomfort related to receiving compassion (Varley et al., 2024), it is plausible these fears may impact awareness of opportunities to receive compassion or reduce the extent to which one allows others to act compassionately toward them. Clarifying this would provide useful information for the development of targeted clinical interventions mitigating negative impacts of attachment insecurity. This may be of particular utility, as while attachment styles are believed to develop in childhood and remain predominantly stable across life (Fraley et al., 2011), capacities for compassion can continue to be cultivated across the lifespan with benefits for wellbeing (Kirby, 2017).

Second, it is unknown how attachment insecurity relates to the experience of compassion in everyday life. Research to date has relied heavily on traditional retrospective-report and laboratory-based observational methods. Given that laboratory settings significantly differ from everyday life settings, it is unclear if previous inferences hold true in everyday life. It is important to clarify this as traditional retrospective-report and laboratory-based methods have limited ecological validity. In other words, the generalizability of such findings to the real world may be limited (Orne, 1962; Bronfenbrenner, 1977). Observing participants in laboratory settings inherently introduces bias as it is impossible for experimenters to entirely reproduce complex natural situations and because laboratory-based experiments are vulnerable to demand characteristics (Orne, 1962). Traditional retrospective-report methods are also susceptible to recall bias. The passing of time between an event of interest such as compassionate action in everyday life and completing a survey related to that event leaves these methods vulnerable to inaccuracy introduced by the fallibility of memory (Kahneman et al., 1982; Klein et al., 1992, 1996, 1997; Marsh and Yeung, 1998; Robinson and Clore, 2002).

Third, it is unknown how attachment may relate to awareness of opportunities for compassion (where ‘opportunities for compassion’ refers to awareness of suffering in the self or others and recognition that there is a chance to act to alleviate that suffering). So far, research has focused on examining attachment-related differences in compassionate actions. Unfortunately, this only captures one aspect of compassion, which has two major components: first, awareness of compassionate opportunities by being sensitive to suffering in oneself and others, and second, a commitment to alleviate and prevent that suffering through action (Gilbert, 2014, p. 19; see also Collins et al., 2006). Research has predominantly focused on how attachment impacts the second of these components. Consequently, it is unclear whether there are attachment-related differences in awareness of compassionate opportunities in addition to differences in compassionate action.

Understanding how attachment relates to awareness of compassionate opportunities rather than just compassionate actions would improve our understanding of the interrelation between attachment and the caregiving behavioral system and provide insight into how attachment influences the processing of information related to compassion (see Shaver and Mikulincer, 2002; Dykas and Cassidy, 2011 for models of attachment-related differences in the processing of information). This would have meaningful implications for clinical intervention. Understanding attachment-related differences in sensitivity to compassionate opportunities would provide greater insight into the stage/s of processing or the component of compassion with which attachment insecurity interferes – awareness of suffering/compassionate opportunities, or engagement in compassionate action. This could inform development and application of clinical interventions. For example, if avoidant individuals are less aware of compassionate opportunities, clinicians may be able to use this information to improve awareness of compassionate opportunities, facilitating greater engagement in compassionate action.

Overall, there is a clear need for ecologically valid research that reviews how attachment relates to the experience of compassion in everyday life, with a focus on both compassionate awareness and action across all three types of compassion: self-compassion, giving compassion to others, and receiving compassion from others. Consequently, our study aimed to investigate whether there are attachment-related differences in compassion-related experiences in everyday life settings, including the likelihood of reporting opportunities to be self-compassionate, to give compassion to others, and to receive compassion from others; in the likelihood of reporting acting self-compassionately, acting compassionately toward others, and allowing others to act compassionately toward oneself; and in the positivity of participants’ emotional responses to these compassionate opportunities.

To address our aim, participants completed a self-report measure of attachment (as attachment is a stable construct, making it less susceptible to issues associated with traditional retrospective-reports; Fraley et al., 2011) and a one-week experience sampling protocol in which participants were asked to complete five short experience sampling surveys each day for seven days. Experience sampling methods (ESM) achieve greater ecological validity than traditional retrospective-reports by asking participants to repeatedly report on everyday behavior and experiences close to the time of occurrence while going about their daily lives (Larson and Csikszentmihalyi, 1983; Csikszentmihalyi and Larson, 1987).

Each experience sampling survey asked participants whether they had observed opportunities to be compassionate to others, self-compassionate, and whether they had the opportunity to receive compassion from others. If participants reported observing compassionate opportunities, they were asked whether they engaged in compassionate actions toward others, toward themselves, or whether others had acted compassionately toward them. Participants also reported the perceived positivity of their emotional experience related to any compassionate opportunities.

We hypothesized that there would be a significant relationship between attachment and propensity to report opportunities for compassion. Given that avoidant individuals fear compassion (Varley et al., 2024) and adaptively suppress negative affective experiences (Shaver and Mikulincer, 2002; Mikulincer and Shaver, 2019), we predicted that greater attachment avoidance would be associated with less likelihood of reporting compassionate opportunities. Given the hypervigilance of anxiously attached individuals to possible signs of threat (Bowlby, 1973; Mikulincer and Shaver, 2003; Gillath et al., 2005; McWilliams and Brodeur, 2016), we predicted that attachment anxiety would be associated with greater likelihood of reporting compassionate opportunities.

We also anticipated a relationship between attachment and engagement in compassionate actions. As above, we predicted that attachment avoidance would have a negative association with the likelihood of engaging in compassionate actions. In contrast, given the tendency for those with anxious attachments to engage in compulsive caregiving, we anticipated that anxiety would have a positive association with the likelihood of engaging in compassionate actions.

Finally, given the dearth of previous research reviewing compassionate behavior in those high in both attachment anxiety and avoidance (sometimes referred to as fearful-avoidant attachment; Bartholomew and Horowitz, 1991), we used exploratory analyses to investigate whether there were any interactions between anxiety, avoidance and likelihood of reporting compassionate opportunities and actions. We also used exploratory analyses to investigate possible attachment-related differences in the positivity of participants’ emotional responses to any compassionate opportunities reported, including whether there were any interaction effects of attachment anxiety and avoidance on their emotional experiences. These analyses were unplanned, and we therefore did not have formal hypotheses. However, previous research utilizing traditional retrospective-report measures indicates that insecure attachment is associated with fear, aversion, and discomfort with compassionate experiences (Varley et al., 2024). It would therefore be plausible that the same pattern might appear in everyday life.

This study was pre-registered and received ethical approval from an institutional review board. Preregistration details, analysis code, and materials are linked here: https://osf.io/tm4cx/?view_only=a7be53dbb2e54a72a5c13d1ba8aeb8fb.

We aimed to investigate attachment-related differences in the likelihood of reporting compassionate opportunities and actions in everyday life and in the positivity of emotional experiences related to compassionate opportunities in everyday life across all three types of compassion: self-compassion, giving compassion, and receiving compassion from others.

As anticipated, those higher in attachment anxiety were more likely to report opportunities for self-compassion, giving compassion, and receiving compassion. Additionally, as expected, those higher in avoidance were less likely to report opportunities to give and receive compassion, but surprisingly, avoidance had no association with reporting opportunities for self-compassion. Further, surprisingly, neither anxiety nor avoidance was associated with the likelihood of giving compassion or allowing others to act compassionately toward oneself. However, there was a significant interaction between anxiety and avoidance on the likelihood of reporting self-compassionate actions such that those with above-average anxiety but below-average avoidance had a lower likelihood of reporting acting self-compassionately. Additionally, attachment anxiety was associated with a less positive emotional experience related to opportunities to receive compassion from others and avoidance was associated with a less positive emotional experience related to opportunities to give compassion to others.

Our findings suggest that attachment anxiety may lead to greater awareness of needs for compassion in everyday life or greater sensitivity to suffering in the self and others. This aligns with research using traditional laboratory-based methods establishing that anxious individuals are more sensitive to emotional expressions than less anxious individuals (Fraley et al., 2006). These findings may result from anxious difficulties down-regulating distress or hypervigilance toward cues that are relevant to assessing and monitoring whether significant others are available and responsive as a means of attempting to maintain proximity to others (Shaver and Mikulincer, 2002; Mikulincer and Shaver, 2003, 2019; Fraley et al., 2006).

Relatedly, Varley et al. (2024) argued that anxious individuals may feel uncomfortable with compassionate experiences because they may interpret the need for compassion in the self or others as a possible threat to one’s ability to maintain proximity to others. Similarly, anxious individuals may be vigilant to cues indicating a need for compassion because these cues may be perceived as potentially signs of threat. This sense of threat may impair one’s ability to act compassionately. This may explain why highly anxious individuals are more likely to notice the need for compassion in themselves and opportunities to receive compassion from others but are not also more likely to act self-compassionately or to report allowing others to act compassionately toward them.

This interpretation may also explain why those high in anxiety but simultaneously low in avoidance are less likely to act self-compassionately. In the absence of the avoidant tendency to suppress negative affective experiences (Mikulincer and Shaver, 2019), the difficulty those high in anxiety have with down-regulating distress may result in prolonged hyperactivation of the attachment system which is thought to interfere with activation of the caregiving system (Gillath et al., 2005), which may then impair ability to act self-compassionately.

These results extend on previous research by more closely pinpointing how differences in attachment relate to differences in compassion-related behaviors. Specifically, our results imply that previous findings that anxiously attached individuals act compassionately less effectively than secure adults (Mikulincer and Shaver, 2017) may not be attributable to lack of awareness of suffering in oneself and others and that previous findings that anxious individuals sometimes engage in compulsive caregiving may be partially attributable to hypervigilance to suffering in others. These findings also support the theory that other issues such as experiencing personal distress in response to suffering may impede the effective provision of compassion (Monin et al., 2010; Vilchinsky et al., 2010), leading these individuals to be less equipped to act self-compassionately or to be soothed by the receipt of compassion from others.

Our research also supports the findings of previous research suggesting that those with greater attachment avoidance are less sensitive to the needs of others (e.g., Kunce and Shaver, 1994; Feeney and Collins, 2001; Braun et al., 2012; Millings et al., 2013; Péloquin et al., 2014), possibly due to discomfort engaging with others’ suffering (Mikulincer and Shaver, 2016; Varley et al., 2024). This may be due to avoidantly attached individuals lacking comfort with closeness and holding negative working models of others (Bartholomew and Horowitz, 1991; Brennan et al., 1998). Further, those with avoidant attachments suppress negative affective states that activate attachment needs (e.g., needs for closeness, support; Mikulincer and Shaver, 2016). This kind of suppression would make it difficult to bring awareness to the suffering of others and recognize opportunities for compassion. Interestingly, however, those with greater avoidance were not less likely to report opportunities for self-compassion. This suggests that while these individuals may suppress negative affective states (Mikulincer and Shaver, 2016), this may occur after having awareness of one’s own suffering. In contrast, our finding that avoidantly attached individuals were less likely to report opportunities to give compassion may suggest that previous findings indicating that avoidant individuals provide less compassion to others may be at least partially attributable to reduced awareness of the suffering of others, rather than an intentional decision to not act compassionately. Further, these individuals report fewer opportunities to receive compassion from others. This may indicate that outward expression of one’s suffering may be inhibited, limiting their ability to access opportunities to receive compassion from others. This aligns with Shaver and Mikulincer’s (2002) assertions and research (Mikulincer and Shaver, 2019) indicating that avoidant individuals use deactivating attachment strategies such as adaptive suppression of thoughts and feelings related to vulnerability or need because of beliefs in others being unavailable or unresponsive.

Overall, our findings in relation to attachment avoidance align with the assertions of Shaver et al. (2010) that if one’s caregiving system develops under unfavorable circumstances, they are likely to become less compassionate with respect to other people’s needs and suffering due to a lack of parental modeling, support, or positive social interactions. However, we argue that, at least in everyday life, this disruption to the development of the caregiving system may result in less awareness of the need for compassion, rather than more conscious choices to not act compassionately.

Our findings also meaningfully extend our understanding of attachment’s relationship with compassion by suggesting that, in everyday life, those with higher anxiety and avoidance are not less likely to act compassionately toward others if they are aware of the opportunity to do so. Indeed, previously identified attachment-related differences in compassionate action may have been attributable to previously unmeasured differences in awareness of compassionate opportunities or sensitivity to suffering. As noted, most previous research only assessed one component of compassion – compassionate action (or, in the case of retrospective-report measures, memories or beliefs about one’s compassionate motivations and actions). Our study extended on previous research by measuring both components of compassion – awareness of compassionate opportunities and compassionate action and by accounting for awareness of compassionate opportunities in our assessment of compassionate action. Thus, this is the first study to review attachment-related differences in compassionate action after accounting for awareness of compassionate opportunities. Consequently, it is possible that we generally observed no differences in compassionate action because attachment-related differences may occur at the level of awareness, rather than the level of action. For example, those with higher avoidance appear to be less sensitive to compassionate opportunities, and thus would be likely to act less compassionately overall.

Thus, our results provide the novel contribution of highlighting the type of compassion-related difficulties insecure individuals experience in everyday settings. Specifically, Gilbert (2014) has argued that effective compassion requires two qualities: first, sensitivity, involving being attuned to and able to accurately understand a person’s signals of need, and second, responsiveness, involving taking useful action, and validating another’s needs and feelings (or one’s own needs and feelings). Earlier researchers also emphasized these same two qualities as underpinning effective caregiving in general (Ainsworth et al., 1978; Reis and Shaver, 1988; Collins et al., 2006). Our results suggest those with insecure attachments likely experience difficulty predominantly with sensitivity.

This was the first study to investigate relationships between attachment and both components of compassion across all three forms of compassion in everyday life. We identified that greater attachment anxiety is associated with greater likelihood of reporting compassionate opportunities across all three forms of compassion while greater avoidance is associated with less. However, the likelihood of engaging in compassionate action was generally not a function of individual differences in attachment, with only one minor exception, with those high in anxiety but low in avoidance being less likely to act self-compassionately.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary material.

The studies involving humans were approved by the University of Queensland Human Research Ethics Committee. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

DV: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Visualization, Writing – original draft, Writing – review & editing. CS: Conceptualization, Methodology, Supervision, Writing – review & editing. JK: Conceptualization, Methodology, Supervision, Writing – review & editing.