BACKGROUND
The concept of earned security has a controversial status in current literature. For many years, researchers have tried to uncover mechanisms that allow people to overcome negative life experiences, and earning security was proposed to be one of them. It suggests the possibility of overcoming a negative relation with a primary attachment figure, thereby asserting flexibility and lack of determinism in caregiver experiences. This article delves into earned security research, reevaluates its applicability, and strives to unveil the role of attachment to alternative attachment figures (AAFs), mentalization, and emotion dysregulation in achieving security.
DOES EARNED SECURITY EVEN EXIST?
Earned security was initially introduced by Pearson et al. (1994) as a distinction from individuals with secure attachment, assessed by the Adult Attachment Interview (George et al., 1985). This term describes individuals who, despite experiencing inadequate care in childhood, exhibit a secure attachment style in adulthood and are capable of coherently discussing their challenging experiences. This denotes a shift from insecure to secure attachment. However, despite attempts to longitudinally the elucidate mechanism behind earned security, the actual levels of parental care during childhood were discovered to be comparable between earned secures and the continuously secure group (Roisman et al., 2002). It was concluded that no transformation in attachment’s internal structure is observed over a lifetime; rather, this group retrospectively perceives parental care as less adequate than it was.
After this publication, researchers lost interest in earning security, and recently it has been discussed only in case studies (Guina, 2016) or qualitative research (Dansby Olufowote et al., 2020). However, Venta et al. (2015) noted that the non-longitudinal nature of changes in attachment should not put the study of the earned secure group to an end. Defining our focus is crucial: is it external reality, where childhood neglect might not have occurred, or mental reality, where an individual recalls inadequate care, impacting their functioning, regardless of facts? Therefore, the so-called retrospectively defined earned secures are individuals exhibiting secure attachment to their parents while evaluating the quality of childhood care from them as low. Hence, at least in these people’s inner world, the parent did not perform well as a caregiver in childhood, but later in life is perceived as someone they can feel secure with. A particular internal reorganization most likely occurred and allowed an individual to benefit from a good internal relationship despite negative evaluation of care early in life. Understanding how this group differs from those with consistent subjective experiences (e.g., low childhood care and insecure adult attachment, or optimal childhood care and secure adult attachment) is important, given the widespread nature of this inconsistency (8-20% of the general population; see Roisman et al., 2002; Saunders et al., 2011; Venta et al., 2015). Addressing these concerns, Venta et al. (2015) proposed replacing “earned secure” with “negative recall subtype of secure attachment”, which does not imply change over a lifetime. We use the term “retrospectively defined earned secure” (abbreviated as “earned secure” – ES) in this paper.
PATHWAYS TO EARNED SECURITY
Little is known of the process through which early childhood experiences are internalized as internal working models and then undergo further changes over life (Konieczny & Cierpiałkowska, 2020; Miljkovitch et al., 2015). It is even more unclear how some people reframe and integrate their adverse caregiver experiences, making them less destabilizing and excruciating. In recent years, researchers have emphasized the significant role of mentalization in reinterpreting traumatic childhood experiences (Chiesa & Fonagy, 2014; Penner et al., 2019). In addition to understanding mental states of others, mentalization embraces understanding, processing, and regulating one’s own mental reality. It proved to act as a mediator between trauma experiences and personality pathology (Ensink et al., 2017; Huang et al., 2020). Mentalization plays a substantial role in reinterpretation of emotional experiences during psychotherapy even when it is not directly focused on mentalizing (Bateman & Fonagy, 2019; Diamond et al., 2014). To our knowledge, there is only one study on mentalizing and earned security. Zaccagnino et al. (2014) discovered that the earned secure group outperformed insecure and continuously secure individuals in terms of mentalization. Robust mentalizing might serve as a protective factor against destabilization due to discrepancies in their inner world, stemming from incoherent childhood relationship experiences and current perceptions of parents. In contrast, insecure individuals struggled to reflect on their negative experiences, resulting in an enduring negative attachment figure perception. The secure group, having fewer chances to “practice” mentalization than earned secure individuals, exhibited relatively lower mentalization. Further exploration of the mentalizing-earned security connection is crucial, revealing its role in reconciling confounding experiences.
The literature also points to emotion regulation as important for internal working models of attachment and for mentalizing, both in developmental and clinical contexts. Studies showed similar levels of emotion regulation in earned secure and continuous secure groups, as well as much poorer emotion regulation in insecure individuals (Brenning & Braet, 2013; Ghiasi et al., 2016; Peng et al., 2021). It is plausible that emotion regulation levels are equally elevated in earned secures and continuous secures, particularly given their comparable performance in overall interpersonal functioning (Saunders et al., 2011). High mentalizing might enhance effective emotion regulation. However, the study of Venta et al. (2015) yielded a contrary result: earned secure adolescents showed a lack of emotional awareness compared to the continuous secure group. This result requires replication, and the difficulties in regulating emotions of earned secures need to be further explored.
In addition to the intrapsychic processes discussed above, situational and interpersonal variables promoting earned security were also proposed. AAFs, people other than parents who played a crucial role in one’s life in childhood (Zaccagnino et al., 2014), provide emotional support, as well as care and instrumental support. This may be considered as corrective emotional experience vital for subsequent changes in an individual’s relational and emotional functioning. Indeed, two studies have confirmed the role of AAFs in earning security, stressing the role of emotional (rather than instrumental) support up to the age of twelve (Saunders et al., 2011; Zaccagnino et al., 2014). Also, psychotherapy is considered as conducive to the development of earned security, especially beyond childhood and adolescence (Guina, 2016; Saunders et al., 2011). Studies show that long-term psychotherapy may lead to changes in attachment representations and to reinterpretation and integration of adverse childhood experiences (Diamond et al., 2014; Kernberg et al., 2008; Levy et al., 2006).
AIMS OF THE STUDY
To our knowledge, this is the first study to investigate emotion dysregulation, mentalization and attachment to AAFs in the context of earned security. Our purpose was threefold: (1) to identify a group of so-called retrospectively defined earned secures – adults who evaluate the level of childhood care from their mother as inadequate but exhibit secure attachment to her; (2) to describe the role of AAFs for this group; we hypothesized that more frequent presence of an AAF and lower levels of attachment insecurity toward that person will be found in earned secure individuals than in secure and insecure ones; and (3) to evaluate the differences in emotional dysregulation and mentalization in these groups; we hypothesized that emotional dysregulation is lowest in earned secure and secure individuals and highest in insecures, and mentalization is highest in secures and earned secures and lowest in the insecure group. In addition, we controlled the time of psychotherapy, education level, and financial status in the family of origin. Besides secure, insecure, and earned secure groups, we also expected to identify a group not described in the literature before, people who had reversed their experience in an opposite way to earned secures. They evaluate the level of care received in childhood from their mother as high but exhibit an insecure attachment to her. We call them “lost secures”. We hypothesized that this group may report even more problems with mentalizing and emotional dysregulation than the insecure group.
PARTICIPANTS AND PROCEDURE
PARTICIPANTS
A total of 272 adult women participated in the study. To increase the heterogeneity of the sample we used two sample sources. The first group was a student sample (N = 219, age 19-33, M = 22.00, SD = 1.75). The second group comprised adults raised in families with one or both parents addicted to alcohol1 (N = 49, age 18-50, M = 33.00, SD = 8.41). Six participants were excluded due to missing data.
Most participants attained secondary education (74%); others had a bachelor’s (19%), or master’s (6%) degree, or vocational education (1%). Family financial background was good (41%), average (39%), very good (13%), below average (6%) and bad (1%), based on subjective reports from participants. Regarding relationship status, over half were married or in a stable relationship (55%), 34% were single during the study, and 11% were involved, but not steadily. About 30% had prior therapy, averaging 12 months (SD = 13.50). Data were collected in 2017 and 2018. Students were approached during lectures. The second sample was gathered through mental health facilities, therapist contact, or purposive sampling from the local population. Trained clinical psychology students carried out the study’s assessment. All individual participants provided informed consent. All procedures were performed in accordance with ethical standards of the institutional research committee.
MEASURES
The Experiences in Close Relationships–Relationship Structures (ECR-RS; Fraley et al., 2011) is a self-reported measure of anxiety about and avoidance of specific attachment figures, here: the mother and AAF. It consists of 9 items for each attachment figure, rated on a 7-point Likert scale. Higher scores indicate higher anxiety and/or higher avoidance, whether low avoidance and low anxiety indicate secure attachment.
The Parental Bonding Instrument (PBI; Parker et al., 1979) is a 25-item self-report measure consisting of two scales, Care and Overprotection experienced from the mother or father. In the present study, we used only the care from the mother scale to identify participants with low and high parental care, as suggested in the previous study (Venta et al., 2015). The respondent is asked to recall interactions with his mother in childhood and early adolescence. Each item is rated on a 4-point Likert scale – the lower the score, the lower the perceived care of the parent.
The Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is a self-report measure of emotional dysregulation consisting of 36 items. The participant answers questions on a 5-point Likert scale from 1 (almost never) to 6 (almost always) – the higher the score, the higher the level of emotion dysregulation. The measure yields scores on five subscales: Nonacceptance of emotional responses (Nonacceptance), Difficulties engaging in goal-directed behavior (Goals), Impulse control difficulties (Impulse), Limited access to emotion regulation strategies (Strategies), Lack of emotional awareness (Awareness) and Lack of emotional clarity (Clarity).
The Mental States Task (MST; Beaulieu-Pelletier et al., 2013) measures mentalization by assessing individual differences in representing/elaborating and being open/modulating towards personal experiences. Emotional arousal was triggered by priming participants with the 3BM card from the Thematic Apperception Test, depicting a character looking sad next to a vague object resembling a gun. They were then asked to write down a story that came to mind in response to the image. Next, using a 7-point Likert scale from 1 (strongly disagree) to 7 (strongly agree), the participants answered 24 questions about their mental state during the story-writing task. The MST captures six mental states, the first two concerned with primitive mentalizing: concrete thinking and low defensive level, then average mentalizing: intermediate defensive level and objective–rational; and mature mentalizing: high defensive level; and reflective thinking. The score on each subscale is scored separately. The total score is derived through an equation using weighted values. The MST has good reliability, with Cronbach’s α coefficient values ranging from .51 to .79 in this study, which is acceptable considering that the subscales are composed of only four items and include different aspects of each mental state (Beaulieu-Pelletier et al., 2021). The reflective thinking scale, however, presents an exception with unacceptable reliability (α = .07 and McDonald’s ω = .35), similar to Tohme et al.’s (2021) findings. As a result, this scale was omitted from the analyses in this study.
RESULTS
DISTRIBUTION TO THE ATTACHMENT GROUPS – IDENTIFYING THE EARNED SECURE GROUP
The earned secure group was identified based on ECR-RS and PBI scores (see Figure 1). ES individuals had secure mother attachment (ECR-RS) but reported low maternal care in childhood and adolescence (PBI) (procedure from Venta et al., 2015). Low care was defined by the PBI cutoff: the authors set 27 as the threshold for low maternal care. Cluster analysis was performed for levels of attachment anxiety and avoidance toward the mother2. The secure group (N = 162) included those with low scores in anxiety and avoidance, while the insecure group (N = 107) comprised individuals with high scores in attachment anxiety, avoidance, or both. In the secure group, we identified individuals with low maternal care (27 points or less on the PBI). The final distribution of the attachment groups was: continuously secure (SEC) n = 122 (46%), continuously insecure (INS) n = 80 (30%), earned secure (ES) n = 37 (14%) and lost secure (LS) n = 26 (10%).
DESCRIPTIVE STATISTICS AND PRELIMINARY ANALYSES
A two-tailed alpha (p < .05) was applied to all statistical tests. The analyses were performed with Jamovi 2.0. Overall, the distribution of MST, DERS, and ECR-RS scores differed significantly from the normal distribution, as indicated by the Shapiro-Wilk W values. The means, standard deviations, zero-order correlations and Cronbach’s α for reliability for all the variables are presented in Table 1.
Table 1
Reliability, means, standard deviations and correlations between study variables (N = 272)
Pearson’s correlations showed weak yet significant negative correlations between mentalizing (mainly MST Low defensive level scale) and some of the DERS scales. Weak negative correlations were observed between mentalizing and attachment anxiety and avoidance, mainly towards the mother. Almost no associations were observed between mentalization and the PBI Care scale. Emotional dysregulation showed positive links with attachment anxiety and avoidance of the mother and the support figure.
DEMOGRAPHICS, RECEIVED PSYCHOTHERAPY, AND ATTACHMENT TO AAFS IN THE STUDY GROUPS
Demographic variables across study groups were compared. We found differences in the financial situation of the participants: the SEC group more frequently reported a “very good” financial state (20% of the SEC group), compared to the other groups (8% in ES and INS, 12% in LS). SEC and LS exhibited more “good” financial status (50% in SEC, 54% in LS) and less “average” status compared to ES and INS groups (30% in SEC, 57% in ES, 45% in INS, 27% in LS; χ2 = 39.80, p < .001). Only participants in the INS group reported a “bad” financial situation (4% of the INS group).
Regarding education, high school degrees were consistent across groups (69-75% for all). A higher proportion of SEC individuals held bachelor’s degrees (26% in SEC, 19% in ES, 14% in INS, 12% in LS), while fewer had master’s degrees (1% in SEC, 19% in LS, 9% in INS, 8% in ES; χ2 = 80.70, p < .001). Furthermore, we observed differences in the number of participants who used psychotherapy (χ2 = 22.00, p < .001). INS and LS used psychotherapy more often than SEC and ES groups (44% in INS, 46% in LS, 25% in ES, and 16% in SEC), although there were no differences in time devoted to therapy (see Table 2). The nonparametric ANOVA test revealed no differences between the groups in terms of the duration of psychotherapy (χ2(3) = 6.19, p = .103).
Table 2
Means, standard deviations, and nonparametric one-way analyses of variance with post hoc analysis for the duration of psychotherapy, attachment to alternative attachment figure, emotional dysregulation and mentalization in four group
Up to 64% of the participants reported the presence of an adult other than parents who was important to them in childhood, up to age 12. Predominantly, this figure was a grandmother (37%), followed by a grandfather (7%), aunt (8%), siblings (8%), and other family members (4%). Teachers (3%) and friends (3.5%) were also mentioned. ES individuals did not differ from other groups in terms of the frequency of recalling an AAF (χ2 = 7.24, p = .065). We did, however, observe significant differences in attachment anxiety to that person, with ES and SEC participants exhibiting lower anxiety than INS and LS individuals (see Table 2). Covariance analysis (ANCOVA) showed that this result cannot be explained by therapy (F(1, 162) = 1.41, p = .237) or education level (F(1, 161) = 2.14, p = .145), but financial status was significant (F(1, 163) = 14.13, p < .001). As for attachment avoidance, SEC individuals scored significantly lower than INS participants, although the ANCOVA revealed the significance of education level (F(1, 161) = 3.93, p = .049), financial status F(1, 163) = 7.87, p = .006) and therapy (F(1, 162) = 10.33, p = .002) as covariates in this model.
EMOTIONAL DYSREGULATION AND MENTALIZATION IN THE STUDY GROUPS
To assess variations in mentalization and emotional dysregulation we conducted nonparametric one-way ANOVA (Kruskal-Wallis test), followed by Dwass-Steel-Critchlow-Fligner pairwise comparisons for independent samples. LS and INS participants demonstrated the most severe deficits across all DERS subscales except Awareness (see Table 2). In the DERS Clarity subscale, the differences between the ES and LS groups and between the SEC and INS groups were statistically significant. In the DERS Strategies subscale, the SEC group showed significantly lower levels of emotional dysregulation than the INS group. In the DERS Nonacceptance subscale and DERS total score, the differences between the SEC group and both the INS and LS groups were statistically significant. Analysis of covariance (ANCOVA) for the DERS total score showed that differences in emotional dysregulation may be related to therapy (F(1, 255) = 9.80, p = .002), but not to level of education (F(1, 255) = 0.87, p = .351) or financial status (F(1, 257) = 1.23, p = .269). ANOVA did not reveal any significant differences among the groups in the DERS Awareness, Impulse, and Goals subscales.
We found significant differences among the attachment groups in terms of mentalizing (see Table 2). The ES and SEC groups scored the lowest on the low defensive level, with post hoc analysis revealing a significant difference between the ES and LS groups and between the SEC group and both the LS and INS groups. The SEC group scored higher on high defensive level and concrete thinking than the LS group; the ES and INS/SEC groups ranked in the middle, with no significant differences between them and the other groups. On the objective–relational scale, differences were observed between high SEC scores and low INS and LS scores. No differences were observed in the intermediate defensive level or the MST total score. The covariance analysis (ANCOVA) showed that the differences in mentalization between the groups are not related to therapy (F(1, 255) = 3.50, p = .062) or education level (F(1, 255) = 0.09, p = .767), but may be related to financial status (F(1, 257) = 5.97, p = .015; for the MST Low defensive level).
The results of the multinomial logistic regression analysis, testing associations between the attachment group as the dependent variable and mentalization and emotion dysregulation as the covariates, are shown in Table 3. The model best suited to the data consisted of the low defensive level of mentalization scales and clarity of emotion dysregulation scales. When covariates were added, the overall model was significant, χ2(6) = 33.40, p < .001, Nagelkerke pseudo-R2 = .07, so the model explains 7% of the variation in the dependent variable. The results showed that people who report a lack of emotional clarity are significantly more likely to be in the INS and LS group than in the SEC group. People who report a low defensive level in mentalizing are significantly more likely to be in the LS group than in the ES group and in the INS and LS group than in the SEC group. In general, both mentalization and emotion dysregulation were related to the distribution to the attachment groups.
Table 3
Summary of multinomial logistic regression analysis for variables predicting four groups of attachment classification
As two of the controlled variables were identified as significant in ANCOVA results, we also built a model with psychotherapy (yes or no) and level of education as factors. When covariates and factors were added, the overall model was also significant, χ2(18) = 65.60, p < .001, and Nagelkerke pseudo-R2 increased to .13, so this model explains more than 13% of the variation of the dependent variable. Regarding psychotherapy, INS and LS individuals are more likely to receive psychotherapy than SEC ones; this was not a significant factor for classification in the ES group.
DISCUSSION
The purpose of the study was to identify individuals who exhibit secure attachment to their mother in adulthood, despite the experience of inadequate care from her in childhood (i.e., retrospectively defined “earned secures”). We tested the significance of attachment to an AAF and psychotherapy duration in this group. We also explored the hypothesis that high mentalizing and low emotion dysregulation may explain why one becomes earned secure while others remain insecure in accordance with their childhood experiences. Our study was cross-sectional in nature, so we cannot claim a change in attachment actually occurred in the course of life. The earned secure is an individual who currently exhibits a secure attachment style to a parent despite perceiving inadequate care during childhood.
In our study, we identified a group of earned secure individuals, which represents 14% of the sample. This result is similar to the rates obtained in other studies, ranging from 8% to 20% (Saunders et al., 2011; Venta et al., 2015; Zaccagnino et al., 2014). The presence of the AAF is reported by most of our sample. Earned secures and secures showed the lowest attachment anxiety levels toward that figure, while differences in attachment avoidance were not significant for the earned secure group. Evidently, the quality of this relationship seems pivotal for corrective emotional experiences, rather than the mere presence of an AAF. A secure emotional bond, often with a grandparent or other family member, devoid of abandonment fear, might be instrumental in reshaping attachment internal models towards greater security, breaking the cycle of intergenerational transmission of insecure attachment styles (Saunders et al., 2011).
We examined mentalization variations among earned secures, secures, lost secures, and insecures. Consistent results were obtained for the mentalization style involving primitive defense mechanisms such as splitting or acting out in response to challenges, which hampers meaningful experience processing. Secures and earned secures exhibited lower levels compared to individuals in the lost secure and insecure groups. Regression analyses confirmed that people who reported less primitive mentalizing are more likely to be in the secure or earned secure group than in the insecure or lost secure group. Blocking threatening mental content with primitive defense mechanisms could detrimentally impede neutralizing negative experiences with caregivers during early life. Nevertheless, it is notable that we did not find any evidence of other mentalizing facets being distinctive to earned secures. This observation contrasts with Zaccagnino et al.’s (2014) study, where they identified the highest reflective function in the earned secure group. These results could be explained by differences between the instruments used to measure mentalization: the self-report vs. rating coding system. As we expected the earned secure group to be similar to the secure group in mentalizing, the differences may be subtle, and more sophisticated methods are needed (e.g., observational tasks or interviews). Furthermore, self-reports capture more of an individual’s beliefs about mentalizing than the actual ability to use it in real life (Jańczak, 2021; Luyten et al., 2019). This may also be the cause of the ceiling effect, when low variation is observed within the highest scores of the mentalization questionnaire. It is also possible that mentalization has limited significance for earned secure individuals, and its role in enhancing or reducing emotional regulation may be crucial here (Ghiasi et al., 2016; Marszał & Górska, 2015; Marszał & Jańczak, 2018).
Our research indicated that individuals in the earned secure and secure groups showed the least difficulty in regulating emotions (with no difference between them), while those in the lost security and insecure groups showed the highest difficulty. Earned secure and secure individuals believe in their ability to influence and modify emotional responses, showing high emotional clarity and unlimited access to emotion regulation strategies. Also, the secure and earned secure groups demonstrated more acceptance of their emotions compared to the insecure and lost secure groups, suggesting that they can manage negative emotions in a more adaptive way. These findings align with the mentalization results above, highlighting the importance of acknowledging one’s emotional states for (earned) security. The relationship between earned security and emotion regulation seems bidirectional. Effective emotion regulation can help reframe negative parent-related experiences, while supportive parental relationships aid emotion regulation during distress. Similar yet reversed mechanisms might operate in the lost secure group. Our study contradicted findings of Venta et al.’s (2015) study on adolescents, where the earned secure group showed higher lack of emotional awareness. It is plausible that it is present only in adolescence as emotion regulation stabilizes, yielding optimal self-awareness of emotional states in adulthood (Marszał & Jańczak, 2018; Penner et al., 2019). Alternatively, this outcome might stem from the DERS Awareness scale’s weakness, indicated by weak latent factor correlations (Fowler et al., 2014). In our study, this scale’s reliability was indeed notably low.
This study contradicted Sounders et al.’s (2011) prior findings that earned secure individuals had more psychotherapy than insecure and secure individuals. It may be due to the general low use of psychotherapy and its short duration in our sample. Research suggests that only long-term psychotherapy can reorganize internal working models, changing experiences of past and present relationships (Diamond et al., 2014; Levy et al., 2006). Unfortunately, we did not gather any data on psychotherapy besides the duration; more details on the modality, course, and cause of treatment could shed more light on this result. However, we did not find evidence that improved parent-child relationship evaluations resulted from psychotherapy-driven insights.
Through our innovative procedure, we identified “lost secures” in our study: individuals who recall high maternal care in childhood but exhibit insecure attachment to her. A disruption has occurred in their relationship with primary attachment figures, as they no longer feel decent care remembered from childhood. Although it was not the aim of this study, our findings revealed that they resemble insecure individuals more than secure ones in seriously impaired emotion regulation and mentalizing.
LIMITATIONS
While the use of less time-consuming tools is essential to examine the earned secure group in clinical and nonclinical samples, the classification of study groups, based on attachment and care score cluster analysis with a defined cut-off point, seems somewhat arbitrary. A longitudinal study or mixed-methods approach, gathering qualitative narrative data about participants’ lifelong parent relationships, would be valuable to validate its credibility. Moreover, to identify the earned secure group in the ECR and PBI data, the alternative approach of latent profile analysis could be used. However, our attempts to identify the ES profile (low attachment avoidance and anxiety, low care from parents) using LPA were unsuccessful. This could stem from the limited sample size, a common hurdle with rare profiles, or if profile variability is low. Additionally, LPA might struggle when variables defining profiles are interdependent. To tackle these challenges, future studies could employ structural equation modelling on a more substantial sample size. Another limitation concerns the Mental State Task, raising concerns about subscales with questionable internal consistency. Further research is necessary to elucidate MST’s correlations with other mentalizing measures (Jańczak, 2021). Also, the study sample warrants consideration: it comprised solely females, lacked exclusion criteria, and omitted any control for variables such as mental disorders, personality pathology levels or trauma experiences at various stages of life. Replicating our findings within a clinical sample, such as individuals with personality disorders exposed to inadequate childhood care, could be valuable.
