Background
Attachment theory provides clinicians and researchers alike a method of examining the impact of early experiences on later adjustment [
1‐
6]. According to attachment theory, early experiences with caregivers are transformed into internal mental representations of attachment during childhood and adolescence [
3,
7‐
10]. Internal models of attachment are theorized to consist of beliefs about the self and others from which rules are derived and used to guide behavior [
4‐
7,
10,
11]. Attachment representations are further believed to affect behavior by influencing the intensity of emotional experience and subsequent attempts at emotional regulation [
3,
8]. In this paper, we utilize the concept of attachment as an organizational construct from which predictions about substance use problems and willingness to seek treatment can be derived [
12].
Attachment theory allows for specific predictions about quality of early experiences with caretakers and the effect of these experiences on future behavior and relationships [
3,
13]. Internal working models of attachment are constructed from repeated interactions with caretakers and are derived from the responsiveness of caretakers during episodes of distress [
4,
5,
7,
14,
15]. These internal representations are hypothesized to act as filters for future relationships and experiences [
1,
7,
9,
13,
16]. For example, responsive and supportive behaviors from caretakers are thought to produce secure attachment representations, which are hypothesized to result in openness to emotional experiences and a willingness to engage in creative and productive emotional regulation. In contrast, unsupportive caretaking (e.g., rejection, neglect) during childhood is thought to be characteristic of dismissing attachment. As a result, individuals classified as dismissing are most commonly characterized as engaging in emotional distancing and greater reliance on the self rather than others. Finally, inconsistent support from caretakers during childhood is most often associated with preoccupied attachment which is thought to produce persistent anxiety towards interpersonal relationships and exaggerated levels of negative affect.
Attachment representations show predictive associations with a wide-range of pathological behavior including personality disorder(s), mood disturbance and psychopathy [
3,
8,
13,
17‐
24]. Empirical support of the association between attachment and problematic substance use is less explored and most studies assess self-reported attachment styles [
25‐
30] rather than internal mental representations of attachment as derived from instruments such as the Adult Attachment Interview [
16]. Studies in which attachment representations (e.g., internal working models of attachment) are assessed typically rely on clinical samples, suffer from limited sample sizes, or provide inconsistent results [
21,
24,
3,
32]. For example, Riggs and Jacobvitz [
32] failed to demonstrate a significant association between substance abuse problems and organized attachment, whereas unresolved attachment surrounding abuse was predictive. In contrast, Rosenstein and Horowitz [
21] found a trend toward higher rates of substance abuse among adolescents classified as dismissing when compared to adolescents classified as preoccupied. Consistent with Rosenstein and Horowitz [
21], Allen, Hauser, and Borman-Spurrell [
31] found a significant positive association between problematic substance use and scales most often attributed to a dismissing state of mind (e.g., derogation of caretakers) and a negative association with scales most often associated with preoccupied attachment (e.g., involving anger). However, Allen et al. [
31] failed to find a significant overall effect of attachment category on substance abuse.
Preliminary analyses of data presented in this paper show significant associations between attachment representations and reports of illicit substance use within a non-clinical sample [
33]. We found significantly higher rates of illicit substance use among individuals classified as dismissing or preoccupied when compared to individuals classified as secure. Although our preliminary analyses provide support for a potential role of attachment in substance use, there were several limitations to our study. First, analyses were limited to the prediction of ever using a substance which combines experimental and problematic users into a single indistinguishable group (i.e., users). Second, variation in inferred childhood experiences within the secure group was ignored (i.e., earned- versus continuous-secure). Although the validity of such a distinction has been questioned, we believe separation of the secure group into continuous versus earned-security is warranted due to differences in rates of psychopathology between the two groups. Earned-security is associated with higher rates of mood disturbance which might increase risk for problematic substance use [
34‐
37]. Finally, the illicit substance use group combined marijuana and non-marijuana substances while alcohol was excluded from the analyses. Therefore, one primary goal of this paper is to examine the association between attachment representations and problematic substance use within a large, non-clinical sample.
A second goal of this project is to examine the association between attachment representations and treatment participation. The presence of substance abuse problems increases the likelihood that professional intervention will be required. In addition to predictions about openness to and intensity of emotional experience, attachment theory can be used to make hypotheses about orientations towards interpersonal relationships and, ultimately, willingness to seek professional intervention [
3]. For example, the experience of supportive and sensitive parenting characteristic of individuals classified as secure is thought to promote a view that others' are available during episodes of distress which will lead to a greater likelihood of seeking support. Experiences of rejection or neglect typically associated with dismissing attachment, on the other hand, is hypothesized to promote feelings of self-reliance and a view that others' are unavailable when distressed. Consequently, individuals classified as dismissing may be less likely to turn to others for assistance. Finally, inconsistent parenting, which is characteristic of preoccupied attachment representations, is thought to produce a hypervigilance towards others' coupled with a continued dissatisfaction with support received. Therefore, individuals classified as preoccupied are expected to report higher rates of seeking professional support due to the unsuccessful impact of intervention.
Although researchers have studied the interplay between attachment representations and therapeutic quality, very little research has been conducted on the impact of attachment on the willingness to seek treatment [
38‐
40]. We are aware of only a single study that examined attachment representations and likelihood to seek treatment. Riggs, Jacobvitz, and Hazen [
11] report significant associations between lifetime history of psychotherapy and attachment representations. Individuals classified as dismissing reported the lowest rates of psychotherapy whereas individuals classified as secure reported the highest. Riggs et al. did not distinguish between earned- and continuous-secure which may contribute to high rates of therapy within the secure group [
11]. Concurrent examination of the association between attachment representations, substance use problems, and willingness to seek treatment will further advance substance abuse interventions.
In summary, attachment theory allows for specific predictions about adaptive and maladaptive behavior in adulthood. We test two primary hypotheses in this paper. Our first hypothesis predicts that individuals classified as dismissing, preoccupied or earned-secure will report higher rates of problematic substance use than individuals classified as continuous-secure. This proposition arises from the role of both inferred childhood experiences and attachment representations in maladaptive emotional regulation [
3,
13]. We predict high rates of lifetime substance abuse/dependence among individuals classified as earned-secure, despite their secure state of mind, due to the influence of inferred negative childhood experiences on substance use. Although ratings on childhood experiences derived from the Adult Attachment Interview [
16] are not veridical with actual experience, we hypothesize that the negative mood proposed to account for the view on childhood experiences will increase the likelihood of problematic substance use [
34,
36]. Problematic substance use among individuals classified as dismissing or preoccupied, on the other hand, is hypothesized to result from an insecure working model of attachment which is thought to promote maladaptive approaches to emotional regulation (e.g., substance abuse). Our second hypothesis predicts different rates of treatment participation as a function of attachment representations. We anticipate low rates of treatment participation by individuals classified as dismissing, despite predicted high rates of substance use problems, due to persistent devaluing of relationships common among this group. We predict high rates of treatment participation among individuals classified as preoccupied, due to hypervigilance towards distress, and earned-security, due to strong valuing of relationships [
1,
32].
Discussion
Attachment theory provides a useful framework from which appropriate interventions into substance use problems can be developed [
1]. Our research questions focused on the association between attachment representations, lifetime prevalence of substance abuse/dependence, and likelihood of participating in treatment. We hypothesized that attachment representations would show predictable associations with problematic substance use and participation in treatment. We further predicted that individuals with certain attachment states of mind would be less likely to seek treatment despite significant problems with substance use.
Our predictions were derived from hypothesized influences of attachment representations on emotional regulation and perceptions about interpersonal relationships [
1,
3,
8,
11]. Specifically, individuals with a dismissing state of mind are believed to view relationships as unimportant and minimize distress. Thus, we predicted that dismissing attachment would be associated with a lower likelihood of seeking treatment despite the presence of substance abuse/dependence [
39,
44,
45]. We hypothesized that preoccupied attachment, on the other hand, would be associated with hypervigilance towards interpersonal relationships and distress. Therefore, higher rates of involvement in treatment and substance abuse/dependence were predicted among individuals classified as preoccupied. Finally, we predicted that secure attachment, regardless of inferred early experiences, would be associated with greater openness towards both emotions and interpersonal relationships resulting in a greater likelihood of seeking treatment. We predicted different rates of substance abuse/dependence between the earned- versus continuous-secure classifications [
34‐
37]. Specifically, we anticipated that individuals classified as earned-secure would report higher rates of substance abuse/dependence than those classified as continuous-secure due to greater negative affect associated with the former group.
The findings were consistent with our predictions. Individuals classified as dismissing reported low rates of participation in treatment despite substantial problems with substance abuse/dependence. Individuals classified as preoccupied or earned-secure reported both high rates of substance abuse/dependence and a greater likelihood of seeking treatment. Finally, individuals classified as continuous-secure reported both low levels of treatment participation and substance use problems.
The pattern of findings for dismissing and earned-secure representations identifies attachment representations as an influential factor in the divergence between the presence of problems and likelihood of receiving treatment [
39,
44,
45]. Both groups had similar inferred negative experiences with caretakers during childhood and similar rates of substance use problems; however, individuals classified as earned-secure were more likely to report seeking treatment. One interpretation from our findings is that attachment representations influence an individual's willingness to seek treatment. Individuals classified as earned-secure, despite their negative inferred experiences with caretakers, continue to value interpersonal relationships. Thus, they are more likely to report seeking treatment. An alternative interpretation, although unsubstantiated with these data, is the potential malleability of attachment representations with appropriate treatment [
21]. Individuals designated as earned-secure in our sample may have been classified as dismissing or preoccupied prior to entering treatment. The earned-secure classification could thus be a consequence of treatment and not a motivating factor in seeking treatment. Finally, the validity of the earned-secure classification continues to be questioned. It is possible that the actual quality of childhood experiences was not accurately reflected by our inferred experience scales due a negative bias accounting for the low ratings [
34‐
37]. This explanation of earned-security would slightly alter the interpretation of our findings attributing substance use problems to the presence of a negative bias and associated mood disturbance rather than to negative childhood experiences per se. Regardless, our findings add further to the apparent qualitative difference with regard to psychopathology between individuals classified as earned- versus continuous secure.
Implications for treatment providers
The findings presented in this paper represent merely a snapshot into the complex role of attachment in the therapeutic process [
1]. At this point, we are only able to speculate about the nuances introduced by attachment representations and the implication of these nuances for the successful treatment of addiction. For example, our findings suggest that a seemingly uniform outcome (e.g., addiction) may result from qualitatively different external and internal experiences thereby requiring modification of interventions to fit "the developing person" [
38]. A continued balance between theory, research, and practice will further advance the development of successful interventions into addiction [
1,
2,
38]. An example of such balance is Flores' [
46] incorporation of attachment theory into a model for the treatment of addiction. He conceptualizes addiction treatment as a "time dependent process" [p. 69, [
46]] in which attachment-related issues alter the focus and nature of intervention across the different stages of recovery. His depiction of addiction as an "attachment disorder" emphasizes the intricate balance that must be achieved when attempting to shift individuals from substance use towards more adaptive methods of emotional regulation. Although it is not always clinically feasible to fully assess attachment related cognitions, interventionists can be trained to watch for hallmark indicators of behaviors associated specific attachment orientations [
1,
2]. Such insight by interventionists will increase the likelihood that addicts will be successful in reducing the rewards of substance use (e.g., negative affect reduction) and promote the development of mentalizing abilities (e.g., reflective functioning) that allow greater insight into the motivations behind use [
46,
47].
Limitations
This study represents a preliminary examination of the association between attachment and treatment. We used a sample of adoptees which limits generalization of the findings. Although the sample was not selected on psychopathology of the adoptee, the sample was balanced on birth parent antisocial and substance use behaviors which increases risk of these disorders in the adoptees. This biological risk increases the odds of psychopathology among the adoptees limiting generalization to a truly random non-clinical sample.
A second limitation involves incomplete information on the purpose and type of treatment, including referral or self-guided solicitation. The questions used to indicate treatment participation were rudimentary at best. Future studies should delineate what type of professional was seen for treatment (e.g., psychiatrist, psychologist, social worker, counselor), the type of treatment implemented (e.g., pharmacological, cognitive behavioral, group), and duration of treatment. Researchers could then explore if certain treatment programs are better matched to certain attachment representations [
1,
46].
A third limitation is the inability to pinpoint timing of treatment which is especially relevant when evaluating earned-security. We utilized lifetime estimates for both substance abuse/dependence and treatment. With adequate data, future studies could examine treatment success in terms of recovery from substance abuse/dependence. Exacerbation of substance misuse could also be observed throughout the therapeutic process. It may be possible that use, or desire to use, may increase as working models of attachment are challenged. This would be especially informative in further understanding the complex interplay between attachment representations and substance use problems.
The final limitation pertains to the omission of the unresolved state of mind surrounding abuse or trauma. Unresolved attachment represents a breakdown of the organized attachment system. Although we could predict elevated substance use problems within this group, we could not make specific hypotheses about help-seeking behavior without taking into account secondary organized classifications (e.g., dismissing, preoccupied, earned- or continuous-secure). Unfortunately, the number of individuals classified as unresolved (n = 39) precluded such analyses.
Competing interests
The author(s) declare that they have no competing interests.