Abstract
There is evidence that within the first year in the foster home, children are capable of forming secure attachment comparable to normative samples. However, less is known about the nature of these newly formed relationships and it has been claimed that they may serve different developmental functions as those to primary caregivers formed in the first year of life. The current study aimed to address this void by applying a behavioral system approach. The sample consisted of 46 children (aged 3–6), 16 of which were currently placed in foster care. Attachment security and dependency were assessed during a home visit using the Attachment Q-Sort. Furthermore, activation of several attachment-related behavioral systems was assessed during a stranger approach in a standardized laboratory setting. Compared to the control group, foster children showed equal levels of attachment security but highly increased dependency scores. Also, during the stranger approach, they showed increased attachment behaviors toward their foster mothers and, at the same time, increased looking behavior toward the stranger. The study points out the relevance of taking into account various attachment-related constructs when investigating formations to new attachment figures in out of home care. It further claims that increased dependency and proximity-seeking scores in foster children are characteristic of the newly formed relationship and a possible indicator of foster children’s adaption. Moreover, this behavior may even be highly beneficial to foster children’s socio-emotional development as it possibly serves important developmental functions.
Introduction
When investigating children’s socio-emotional adaption during the transition to foster care, attachment theory provides a strong theoretical framework for several reasons. By providing foster children with a safe place to grow, they get the opportunity to benefit from continuous, warm, and supportive relationships, which may potentially buffer previous negative experiences (Bowlby, 1988; Sroufe, Egeland, Carlson, & Collins, 2005). Even though, at the time of placement, foster children show to be significantly less securely attached than their counterparts, longitudinal data suggest a significant catchup during the first year of placement (Gabler et al., 2014; Lang et al., 2016). More precisely, Gabler and colleagues (2014) found an increase in attachment security during the first 6 months within the foster home which was followed by a further increase until 12 months after placement (Lang et al., 2016) with foster parent’s sensitivity and an authoritative parenting style decisively contributing to heightened security. Also, several other studies found foster children to be capable of forming secure attachments to their foster parents (Cole, 2005; Jacobsen, Ivarsson, Wentzel-Larsen, Smith, & Moe, 2014; Oosterman & Schuengel, 2008; Ponciano, 2010; Stovall-McClough & Dozier, 2004). This is also supported by empirical data (van den Dries, Juffer, van IJzendoorn, & Bakermans-Kranenburg, 2009). However, there is reason to assume that the nature of these relationships is not necessarily fully comparable with that of secure attachment relationships formed within the biological family under healthy circumstances.
Before being placed into a foster family, foster children often have experienced physical and emotional maltreatment or neglect in their biological families (Minnis, Everett, Pelosi, Dunn, & Knapp, 2006; Takayama, Wolfe, & Coulter, 1998), which is associated with insecure working models of attachment (for a review, see Baer & Martinez, 2006). These pre-experiences may challenge the formation of new secure attachments (Dozier, Stovall, Albus, & Bates, 2001; Stovall & Dozier, 2000). Also, foster children carry the burden of a sudden loss of continuity. In particular, the loss of a primary caregiver has shown to have debiliating effects on development in humans and non-human primates (Bowlby, 1973; Cirulli, Berry, & Alleva, 2003; Suomi, Collins, Harlow, & Ruppenthal, 1976), which implies a considerable risk for maladaption in foster children. Taken together, attachment relationships in out-of-home care are often based on uncertain grounds and thus need further investigation.
Indeed, regarding the nature of attachment relationships to new caregivers, there seem to be some major differences. For example, foster children, as well as adoptees, are more likely to show disorganized attachment patterns (van den Dries et al., 2009). Also, there are numerous studies suggesting that foster children may form secure attachment relationships with their foster parents, but, nevertheless, maintain symptoms of disinhibited attachment behavior, which seem to be more resistant to environmental change (Chisholm, 1998; O’Connor, Marvin, Rutter, Olrick, & Britner, 2003; Zimmermann, 2015). In sum, this suggests major aberrations within certain behavioral systems that organize foster children’s social and attachment behavior.
Still, Lyons-Ruth (2015) further regards attachment as a powerful framework to explain nonnormative social behaviors in children with a history of pathogenic care. However, she questioned whether the newly formed attachment relationships to new caregivers really “serve the same developmental functions in organizing the child’s emotional, behavioral, and physiological systems” (Lyons-Ruth, 2015, p. 226). Thereby, she emphasizes the need for a more profound understanding of the nature of attachment-relationships formed after the first 2 years of life. The current research responds to this void in the literature by applying a behavioral system approach that can inform our understanding of foster-mother–child relationships beyond attachment security.
Behavioral systems associated with attachment
Taking an ethological perspective, Bowlby (1969, 1982) theorized attachment as a function of several inherent behavioral systems, that when activated elicit certain behaviors leading to a specific predictable outcome. For example, behaviors that ensure closeness between caregiver and child from birth on (e.g., infant cry) are governed by the attachment behavior system. Strongly connected to the attachment behavior system is the construct of dependency, yet it needs to be distinguished from attachment. Inarguably, infants highly depend on the close presence of an “individual who is conceived as better able to cope with the world” (Bowlby, 1988, p. 27). Thus, at early stages, attachment behaviors and so-called dependency behaviors (e.g., clinging to the caregiver) essentially overlap. However, while the concept of dependency is entirely related to overt behavior (Ainsworth, 1969), attachment security functions in substantially different ways (Ainsworth, 1969, 1972). Importantly, the infant is dependent on the caregiver (and thus shows increased attachment behavior) even before becoming attached to him/her and, later, he/she can be attached to a caregiver even though not depending on his/her presence in a given situation (Sroufe, Fox, & Pancake, 1983; Sroufe & Waters, 1977).
Applying an organizational perspective (Sroufe & Waters, 1977), it can be assumed—and evidence confirms—that children’s need for physical proximity decreases with age and overt attachment behaviors become less frequent (e.g., Marvin & Greenberg, 1982). By then, internal representations of—optimally—a secure attachment relationship provide the basis for developmental tasks like the endurance of prolonged separations from the caregiver, and thus facilitate autonomous functioning (also see Wong et al., 2011, p. 490). Thus, it is suggested that children, who were able to effectively depend on their caregiver and develop a secure attachment relationship with him/her early in life, act more self-confident and autonomously in later stages of development. Indeed, Sroufe, Fox, and Pancake (1983) found that children who were insecurely attached to their caregiver at 12 and 18 months were described as highly dependent on their teacher in preschool. Despite its informative value, during the last decades, the investigation of dependency within caregiver–child dyads has been widely neglected.
Beside attachment behaviors, those behaviors governed by the wary behavior system, which gets activated by novel situations and persons, also have the effect of maintaining proximity to the caregiver. Thus, these two systems primarily serve the biological function of protection (Marvin & Britner, 2008, p. 272). Still, actively venturing forth and learning about the environment becomes increasingly important with age (Ainsworth & Bell, 1970). The behavior systems promoting this process are the exploratory behavior system and, as Ainsworth (1990) and Marvin (1977) suggested, the sociable behavior system. Again, the phenomenon that with increasing age feelings of closeness are no longer based on physical contact but on “internal states” (Cicchetti, Cummings, Greenberg, & Marvin, 1990, p. 15) allows the toddler to spend more time away from the caregiver and explore the environment.
Importantly, Marvin and Britner (2008) stated that systems are continuously in an “equilibration process,” where activation of one system leads to the termination of an antithetical system. While this interplay is best explained by the concept of the attachment–exploration balance (Ainsworth, 1967; Ainsworth & Bell, 1970; Bowlby, 1988), its complexity becomes even more evident when acknowledging different forms of behavioral systems toward strangers (i.e., the sociability system and the wary system) (Sroufe, 1977). Greenberg and Marvin (1982) were the first to describe preschool-aged children’s reactions to a friendly stranger in terms of qualitative patterns that result from the interplay between the different behavioral systems. In their observational study, they have found that in most 2- to 4-year-old children, an initial confrontation with the stranger deactivated the exploratory system for a short time where children sought physical proximity to their caregiver. Interestingly, they found a developmental shift occurring at 3 years of age referred to as a “developmental coupling of…systems” (p. 489), which decreases between the wary and the attachment system and increases between the wary and sociable system. Thus, in normative development, children may still activate the wariness system when being confronted with a stranger but progressively tend to be less dependent on physical contact with the caregiver.
In sum, the dynamic balance among these behavioral systems changes with development in a typical manner; however, in dissimilar caregiving environments, different patterns of “equilibrated organization” (Marvin & Britner, 2008, p. 273) may arise. As suggested by O’Connor, Marvin, Rutter, Olrick, and Britner (2003), a lack of a warm and predictable relationship “may have impaired the child’s ability to develop (interaction-based) strategies for regulating his/her arousal level and for the activation and termination of attachment, exploratory, wary and sociable behavior” (p. 34). Still, research with high-risk samples mostly focuses on either attachment security or children’s sociable behavior toward strangers without taking into account the balance and functions of different behavioral systems that are in play during social interactions.
Study aims
The first major aim of the current study was to explore how foster children 1 year after placement differ from an age-matched control sample with regard to both constructs, first, attachment security, and second, dependency. Referring to the longitudinal finding of an increase in attachment security with the same overall sample (e.g., Lang, 2014), it is hypothesized that 1 year after placement foster children will not differ from a control sample regarding attachment security. However, we assume that the nature of the attachment relationship between foster mother and foster child differs from one that has been established in the biological family under typical circumstances. More precisely, since the quality of early caregiving relationships is negatively related to dependency in the preschool years (Sroufe et al., 1983), we expected that despite an improvement in attachment security, foster children would still show increased dependency as compared to control children.
Second, provided that foster children show increased dependency toward their foster mother 1 year after placement, we aimed to determine whether these aberrations within the foster mother–child relationship would also show in a standardized laboratory assessment. Therefore, we assessed how foster children as compared to control children regulate their behavior when gradually being approached by a stranger in the caregiver’s presence. By the stranger approach and the novel situation, we aimed to activate children’s attachment behavior system. Along with our first expectation on increased dependency in foster children, we expected foster children as compared to control children to show increased attachment behavior toward the foster mother during the procedure. Furthermore, with the stranger providing toys within the interaction task, we also aimed to activate the exploratory system and the sociability or wariness system. The interplay of these systems is assumed to reflect an inner organizational model resulting from caregiving experiences (Ainsworth & Bell, 1970). However, severe disruptions during the formation of attachment relationships are likely to cause major alterations, making the “interrelatedness” of behavioral systems less predictable (O’Connor et al., 2003, p. 34). Thus, we further aimed to investigate differences between foster children and a control group regarding activation of these behavioral systems.
Method
Participants
The total sample consisted of 46 preschool-aged children (23 males/23 females), 16 of which were living in foster care (9 males/7 females). The remaining 30 control children were living in their biological family (14 males/16 females). Two more control children were tested, but not included in the final sample, because of missing tests. Children’s age at the laboratory assessment ranged from 3 years to 6 years (M foster children = 54.36, SD = 11.66; M control children = 55.31, SD = 10.07; p = ns). At the home assessment, foster children were younger than control children (M foster children = 46.21, SD = 13.97; M control children = 57.30, SD = 10.10; p = .003), as children’s age was matched with regard to the laboratory assessment. Age effects were tested preliminary and subsequently controlled for, when indicated.
The foster care group forms a subsample of foster children, who took part in a preceding 1-year longitudinal assessment (see Bovenschen & Spangler, 2013) and met our age criteria (3–6 years). Eight of the 24 foster children from the regional subsample did not attend, as foster mothers described them to be too impulsive (n = 1), too emotionally unstable (n = 1) or indicated other unspecified personal reasons (n = 6). Re-recruited foster children did not significantly differ from the rest of the sample regarding age at placement, attachment security, dependency, signs of disinhibited engagement disorder as reported by the foster mother, and problem behavior, 1 ps > .05 (Table 1).
Dropout analysis for foster children’s age at placement, attachment security, dependency, disinhibited attachment, and problem behavior.
aScores represent z-values.
bScores of Disturbances of Attachment Interview—Disinhibited scale.
c t-Values total score of Child Behavior Checklist.
Foster children’s characteristics
At the time of the laboratory assessment, the average time children had spent in the current foster home was 21 months (range: 15–30 months, M = 21.09, SD = 4.61). Reasons for placement in foster care were manifold, and several children had experienced multiple forms of endangerment of their welfare. These included maltreatment (n = 15), violation of duty of supervision (n = 13), neglect (n = 11), parents’ mental illness (n = 11), physical maltreatment (n = 5), sexual abuse (n = 2), and parent’s imprisonment (n = 2). One child was given into care voluntarily. 12 of 16 children have had experienced at least one previous placement.
Family characteristics
Seventy-five percentage of participating foster mothers and 66% of mothers in the control group indicated to be the child’s primary caretaker, while the rest were sharing caretaking responsibilities with their partners. In less than half of the foster families (43.7%) and control families (40%), mothers were working in a full- or part-time position. Demographic data indicated that foster mothers on average were significantly older than mothers in the control group, t(44) = −5.84, p < .001, and, that control families were more likely to have an academic background (53% vs. 13%). These differences are assumed to represent common characteristics of the two family types.
Design and overall procedure
An ethical review of the study protocol was approved by the ethic commission of the German Association for Psychology (DGPs). All caregivers gave written informed consent after they were informed about the procedure and study aims. They were further informed that their participation was voluntary and could be terminated at any point. One home and one laboratory assessment was conducted. Children’s attachment security and dependency was assessed at a videotaped 2.5 hr long semi-structured home visit. At the lab visit, attachment behaviors toward caregiver, exploration, and social behaviors toward strangers were assessed in a standardized stranger-approach referred to as the waiting room procedure.
For the current study, data on foster children’s attachment security and dependency were retrieved from the longitudinal study’s final home assessment, which took place 1 year after placement in the foster home (M = 12.62, SD = 0.91). About 8 months later, the laboratory assessment took place (M = 8.15, SD = 4.24). Notably, in control children, the home assessment was conducted close to the laboratory visit (M = 1.59 months, SD = 0.84). This issue is addressed in the “Limitations” section.
Measures
Attachment Q-Sort
The observer Attachment Q-Sort (AQS; Waters & Deane, 1985, German version: Schölmerich & Leyendecker, 1999) has shown good psychometric properties and is a widely used measure for assessing attachment in children up to 70 months (van IJzendoorn, Vereijken, Bakermans-Kranenburg, & Riksen-Walraven, 2004). It also allows for the rating of a child’s dependency. The AQS includes 90 items printed on cards, which are sorted into nine piles categorizing the child’s behavior from “less characteristic” to “most characteristic.” Subsequently, the child’s Q-Sort is correlated with criterion sorts developed by experts (see Waters & Deane, 1985; Waters, Vaughn, Posada, & Kondo-ikemura, 1995). The correlation results in a security score ranging from very insecure (−1) to very secure (+1) and a dependency score ranging from very independent (−1) to very dependent (+1). In the present study, interobserver reliability was high, r = .70. Composites of both observers’ security and dependency scores were calculated. Correlation coefficients were converted into Fisher’s z (Field, 2009).
Waiting room procedure
Behavioral assessment
To observe children’s immediate reactions to a gradual stranger approach in a quasi-natural setting, a standardized 10-min long waiting room situation was set up. It was adapted from studies, which included children adopted from institutions and out of foster care (Bruce, Tarullo, & Gunnar, 2009; Lawler, Hostinar, Mliner, & Gunnar, 2014). Furthermore, as unfamiliar situations are expected to activate the attachment system (Ainsworth & Wittig, 1969), this behavioral procedure seemed suitable to illustrate the absence or presence of children’s attachment behaviors toward the caregiver.
Caregiver and child were led to a room with five chairs that were arranged forming an “L.” The caregiver sat on the outer left and signaled to be busy with paperwork. The child was to sit on the chair in the middle corner and was handed a picture book. The session started when the stranger entered the room and sat on the outer right chair. The first phase (introduction), in which the stranger greeted the child lasted 1.30 min. In the second phase (toy offering), the stranger provided some toys, verbally offered the child to play with them, and sat back on the chair acting passively. After 4 min, the third phase started (invitation to play). Here, the stranger got out more toys, sat on the floor, and actively invited the child to play together for 4 more minutes. The entire session was videotaped with two cameras positioned opposite to each other.
Behavioral coding
Coding was made with regard to children’s attachment behavior toward the caregiver, exploratory behavior, and looking behavior toward the stranger using a lab-based rating system with scores ranging from 0–3 for each of the three behavioral sets. To increase precision of coding, all phases were separated into 30 s segments. As for attachment behavior, a score of 0 indicates that the child did not show any attachment-related behavior (i.e., seeking proximity to caregiver, back-checking) while a score of 3 was given, when the child was seeking and/or maintaining proximity to the caregiver for more than 50% of the time. As for exploratory behavior, a score of 3 indicates continuous exploration of toys while 0 was given when he/she did not play for at least 80% of the time. Finally, as for looking behavior toward stranger: if the child did not even once look in the stranger’s direction, a 0 score was given while 3 indicates that he/she was looking at the stranger the entire 30 s or was continuously taking short glances at her. One pair of coders coded the attachment–exploration balance after intense training. Post hoc inter-rater reliability of 30% of videos led to averaged KW s between .76 and .96. For each of the three behavioral sets (attachment behavior, exploratory behavior, looking behavior towards stranger), scores given for the 30 second segments were averaged per phase (introduction, toy offering, invitation to play).
Statistical considerations
To compare scores in foster children and control children, t-tests and analyses of variance (ANOVAs) were conducted. The assumption of equality of variances was not violated in any analyses. For repeated measure analyses of variances (RM-ANOVAs), Greenhouse Geisser correction was applied when necessary. Post hoc tests were done using t-tests for independent samples and pairwise Bonferroni corrected comparisons. Kolmogorov–Smirnov tests revealed that the assumption of normal distributions was not given throughout. Thus, for correlations, Spearman’s rank coefficient was used. Where possible, nonparametric tests (Mann U-Whitney, Friedman’s test) were run additionally and confirmed significant effects. Extreme values (n = 3) were adapted by changing their value to the minimum/maximum value within the acceptable range minus/plus one measuring unit.
Results
Preliminary analysis showed a zero-correlation between attachment security and dependency, r = −.02, ns, while age at the home visit was negatively correlated with dependency, r = −.52, p < .001, but not with attachment security, r = .17, ns.
Differences between foster and control children regarding attachment security and dependency
We first addressed the question, whether foster children and control children differed with regard to attachment security and dependency.
As expected, comparison of AQS security scores showed that foster children (M = 0.30, SD = 0.21) did not differ from control children (M = 0.29, SD = 0.21), t(44) = −2.22, p = n.s. However, regarding dependency, we found that foster children’s dependency scores ranged between −.26 and .47 (M = 0.14, SD = 0.23) while control children only exhibited negative scores ranging from −.48 to −.14 (M = −0.17, SD = 0.17). This difference was significant, t(44) = −5.13, p = .000, with a large effect size, d = 1.6 (Cohen, 1988). Considering that younger children were generally more likely to show dependent behavior, we conducted an analysis of covariance regarding dependency including age as a covariate. This did not change the results with respect to the level of significances.
Differences between foster and control children’s interactional behavior toward caregiver and stranger
Next, we tested whether foster children and control children differed with regard to their behavioral responses to the stranger approach in the waiting room at the laboratory assessment.
Regarding intercorrelations between measures, Table 2 presents a positive association between attachment behavior and looking behavior toward the stranger while both of these variables were negatively related to exploratory behavior in all children. None of the variables significantly correlated with age. Furthermore, they were not related to attachment security. However, there was a high positive correlation between overt attachment behaviors as well as looking behavior toward the stranger and children’s dependency scores. These correlations remained essentially the same after controlling for child age.
Correlations with age and between behavior variables assessed in the waiting room procedure and with the AQS.
Note. AQS = Attachment Q-Sort; Spearman-Rho. N = 46.
aAcross all phases of the waiting room procedure.
***p < .001; **p < .01; *p ≤ .05.
For each of the three behavioral variables (attachment behavior toward caregiver, looking behavior toward stranger, and exploratory behavior), a repeated measures ANOVA was conducted including phase (introduction, toy offering, invitation to play together) as the within-subjects factor and group (foster, control) as the between-subjects factor. Figure 1 illustrates behavioral differences between foster and control children unfolding over the procedure’s three phases.
Attachment behavior toward caregiver, looking behavior toward stranger, and exploratory behavior throughout the waiting room procedure by group.
For attachment behavior toward the caregiver, there were significant main effects for phase, F(1.74, 76.71) = 28.95, ε = .87, p = .000,
Regarding children’s looking behavior toward the stranger, the repeated measure analysis revealed a main effect for phase, F(2, 88) = 12.81, p = .000,
Regarding exploratory behavior during the stranger approach, the repeated measures ANOVA revealed a significant main effect for phase, F(1.46, 64.43) = 10.81, ε = .73, p = .000,
Summing up, during the waiting room procedure, both foster and control children exhibited a dynamic attachment exploration balance with heightened attachment behavior during introduction and toy offering followed by a decrease after play invitation and the complementary pattern of low exploration before invitation followed by an increase afterward. However, there also were two main differences between foster children and control children. Specifically, foster children generally showed a heightened level of attachment behavior to their caregiver and they showed more looking behavior toward the stranger during play offering.
Discussion
The current study aimed to broaden our understanding of attachment in foster children. Thus, we were not only looking at attachment security within the caregiver–child relationship, but also focused on dependency. We further investigated whether during a caregiver–stranger interaction foster children would show specific behavioral patterns regarding attachment behavior to caregiver (i.e., proximity seeking), exploratory behavior, and stranger sociability/wariness.
Regarding attachment security and dependency, we found empirical evidence for expected conceptual differences. More precisely, attachment security and dependency were not related to each other. This indicates a clear distinction between the two constructs theorized by Ainsworth in the early years of attachment research (Ainsworth, 1969, 1972). Also, we found that dependency, but not attachment security, was negatively correlated with age. This goes along well with a developmental model, where dependent behaviors are more common in younger children as they need closer proximity of the caregiver as a base from which to explore the environment while with age, children establish an internal working model enabling them to move further away and benefit from more learning opportunities.
With regard to attachment security in children in out of home care, previous research suggests that they are able to establish a secure relationship with their new caregiver (see van den Dries et al., 2009). In consistency, longitudinal data including our subsample suggest that attachment security in foster children increases during the first year of placement (Gabler et al., 2014; Lang et al., 2016).
Thus, rather unsurprisingly, by 1 year in foster care, we found foster children to not differ from a control group with regard to attachment security. Importantly, they still clearly differed with respect to their dependency within the caregiver–child relationship. More precisely, even after controlling for age, foster children showed more dependent behaviors than control children. This phenomenon has been found during the home observation but also the standardized laboratory procedure (see below).
Our analyses confirmed that the waiting room procedure was appropriate for activation and deactivation of the attachment behavior system, the exploratory behavior system as well as the sociability system, as the three behavioral variables systematically varied as a function of phase. More precisely, attachment behaviors were elicited during the stranger’s approach (introduction, passive toy offering). The stranger’s invitation to play together resulted in an increase of exploration behavior accompanied by a reduction of attachment behavior in all children. These main effects demonstrate that the attachment–exploration balance was well functioning both in the control and the foster children.
Importantly, we also found two major behavioral differences between the groups during the laboratory session: first, increased attachment behavior toward the caregiver and, second, increased looking behavior toward the stranger.
More precisely, our findings show that during the laboratory session, foster children were as likely as control children to explore the environment. However, they needed more proximity and/or repeated reassurance of the caregiver’s presence to establish a sufficient extent of security. This goes along with our finding that foster children show normal levels of attachment security but at the same time, their level of dependency was increased. Thus, we found that the relationship shows specific characteristics despite the fact that it can be classified as normatively secure. To explain increased dependency and need for reassurance and physical proximity in foster children, we suggest that the formation of this close attachment relationship follows similar stages than the formation of the infant’s attachment to the primary caregiver in the first year of life. Due to human nature, infants depend on their caregiver from early on. With age, an inner working model of the caregiver is formed, enabling children to spend more time away from him/her. Consequently, in normative development, it is expected that dependent behaviors may occur less frequent with age as children finally develop into more self-reliant human beings. This interpretation is supported by the negative correlation between dependency and age we found in our study. Interestingly, in our study, compared to a control group, foster children’s dependent behaviors toward the foster mother were still increased 1 year after placement and the effect remained after controlling for age. Here, it could be argued that foster children in our sample may still be in an early stage within the formation of the new attachment relationship. Thus, their attachment-related behavior resembles that of younger children that have not yet internalized an inner working model based on experiences with a specific attachment figure. This internalization may be especially challenging and time-consuming, as it possibly means a cognitive restructuring of an early model that derived from previous pathogenic experiences. Taking into account these early experiences, it is also possible that certain behavioral systems were not able to develop normatively and thus remained on a premature stage. In this line, the foster mother–child relationship may serve the function of providing a secure space for “post-ripening” of important socio-emotional competencies. Taken together, we argue that dependency and increased attachment behaviors like proximity seeking in foster children are special characteristics inherent to the newly formed attachment relationship that is yet to be established on the unstable bases of early adverse relational experiences. Furthermore, they may be an indicator of foster children’s adaption to the new context and serve an important developmental function within the socio-emotional domain. Yert, more long term assessments are needed to confirm this assumption.
Regarding looking behavior toward the stranger, we found control children to display a low and invariable level across all phases, while foster children showed a different pattern. Especially during toy offering the latter showed enhanced looking behavior directed toward the stranger, who was instructed to not engage with the child during this phase. There are at least two different explanations for increased looking to stranger in foster children: on the one hand, it may be interpreted as a sign of vigilance developed as an outcome of early experiences within a rather unpredictable social environment. If this interpretation is true, increased looking behavior toward the stranger can be seen as an attempt to monitor her actions and detect possible threat. On the other hand, referring to the concept of disinhibited social engagement toward strangers (see Zimmermann, 2015), increased looking at the stranger may be interpreted as heightened interest in the stranger’s behavior and as an attempt to engage with her. Notably, these explanations do not have to be mutually exclusive as both emphasize foster children’s attention to strangers, which, for example, has also found to be evident on a neurophysiological level (Kungl, Bovenschen, & Spangler, 2017). With regard to our findings on increased attachment behaviors, it may be possible that the stranger elicits both, wariness and sociability behaviors. Being in the process to integrate oneself into a broader social network in an appropriate way may elicit apparently contradictive behaviors simultaneously. Indeed, the sociability and the wariness system have shown to be coupled in 2-year old normative developing children as well (Marvin & Greenberg, 1982). Thus, foster children’s behavior may represent some kind of delay in the attunement of behavioral systems.
Finally, when looking at intercorrelations between the key constructs attachment security and dependency and foster children’s behavior in the standardized laboratory procedure, we found that attachment security was not related to any of the behavioral variables. This confirms that attachment security is a complex construct that describes the balance of certain behaviors rather than their quantity. Notably, there were no correlations between attachment security and attachment behaviors, which, however, were strongly associated with dependency toward the caregiver. Indeed, increased attachment behaviors (i.e., proximity seeking) are inherent to the construct of dependency. Thus, attachment behavior in this age group is no sign of security, rather is it an overt behavioral pattern characterizing a child that depends on the caregiver’s proximity, hence aiming to establish and maintain it. Also, there was no association between dependency and exploratory behavior. Furthermore, dependency (and attachment behavior) was positively related to looking behavior toward the stranger. Thus, it is suggested that in our sample, the attachment behavior system and the stranger sociability system/wariness system were activated simultaneously, a phenomenon that has already been found in younger preschool children (Greenberg & Marvin, 1982).
Limitations
The study has several limitations. First, we included 46 children, but the number of foster children was relatively small (n = 16). This uneven distribution was due to the study design, as this study complemented a preceeding longitudinal assessment. As foster children form a highly vulnerable sample, a substantial number of foster parents refused to attend any additional measuring beyond the 1-year assessment. In a drop-out analysis, however, we showed that the remaining sample of foster children did not differ from the original sample, neither in age, nor in any of the central study variables, nor in mental health status. In our statistical analyses, we paid additional attention to the fact that data from small samples are especially sensitive to extreme values and that some distributions were nonparametric. Consequently, we expect significant results not to be due to statistical artifacts.
However, it cannot be excluded that foster families’ characteristics—and thus, foster children’s recent caregiving experiences—might differ from those of biological control families. In fact, in the current sample, many foster mothers already brought up own children, and thus, had a different parenting background than control mothers. Furthermore, foster and control families significantly differed regarding their educational background. Hence, children may not only have experienced different caregiving environments during the early years, but also during the previous year. However, Lang (2014) compared foster parents and mothers from the current control sample regarding parenting behaviors, but did not find any differences by 1 year after placement. Thus, parenting behaviors are not assumed to account for group differences in the current study.
Further methodological limitations refer to the fact, that—with regard to the laboratory assessment—AQS data (attachment security and dependency) in foster children were collected a few months earlier than in control children. Thus, it cannot be ruled out that AQS scores in foster children may have changed by the laboratory assessment. Still, continuity of scores was assumed for the following reasons: The longitudinal assessment of the overall foster sample (n = 55) showed that attachment security only changed during the initial 6 months of placement and then remained stable (e.g., Lang et al., 2016). Also, in normative samples—once the relationship is established—attachment security appears to be a stable marker of the caregiver–child relationship (e.g., Wartner, Grossmann, Fremmer-Bombik, & Suess, 1994). Still, further research on the development of attachment relationships in foster care is needed for more definite conclusions.
Conclusion
Keeping in mind that by theory, the organizational structure of the attachment system develops along a “developmental sequence from basic attachment behaviors […] to a representational organization later on” (Spangler & Zimmermann, 1999, p. 270). It may be that after 1 year in the foster home, children have not yet internalized the foster mother as a secure base. In normative development as the caregiver’s actions become more predictable, the child needs to refer to him/her less through bodily proximity. In other words, it has developed an enhanced “theory of attachment” including detailed knowledge about the caregivers support (Delius, Bovenschen, & Spangler, 2008). At later developmental stages, heightened attachment behaviors (as we found in foster children) may thus be indicative of a less mature way to maintain psychological security within novel situations.
Finally, on the background of early adverse experiences, dependency and increased attachment behaviors may serve different developmental functions than in normative development and could be understood as a sign of current behavioral reorganization (see Sroufe, 1979). Interestingly, Poitras, Tarabulsy, St-Pierre, and Varela Pulido (2015) argue that increased dependency in foster children may be a favorable indicator of a new “attachment in the making,” and thus should be acknowledged as possibly nourishing the bond between foster child and caregiver (Poitras, Tarabulsy, St-Pierre, & Varela Pulido, 2015). Indeed, these behaviors may facilitate the strengthening of the newly formed relationship by making the foster parent feel more related, and finally enable the child to catch up on experiences of an intimate relationship to a reliable attachment figure.
With the current study, we reveal a multifaceted picture of the newly formed attachment relationship between preschool-aged foster children and their foster parent. The study provides further evidence that nonnormative experiences during the attunement of behavioral systems may lead to aberrations regarding the interplay of behavioral systems in terms of their regulatory functions (see O’Connor et al., 2003, p. 34) and broadens our understanding of attachment in foster children. By identifying specific behavioral patterns and providing a theoretical framework, this study can inform practical work with foster families by helping them to make meaning of their foster child’s behavior, interpret it in an resource-oriented manner and respond to it appropriately. Finally, such adequate support may further help to reduce family stress and facilitate foster parents’ sensitivity (see Gabler et al., 2018).
Footnotes
Acknowledgements
First of all we want to acknowledge the support from our funders, which set the financial base for this study. We also thank our students for their help with data collection and especially emphasize their sensitivity when dealing with the young children. We further thank Karine Poitras for sharing some fruitful thoughts on dependency in children in out-of-home care. Of course we couldn’t have done the study without the participating families and thus want to thank them for their time and willingness to contribute to research in the field.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Staedtler Stiftung and the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG, grant number SP 312/19-1 and NO 891/1-1).
