For more than 20 years, our attachment research teams at Western Michigan University (WMU) have been using Bowlby’s attachment theory to examine important psychotherapy process and outcome variables. What have we found? Generally speaking, client and therapist attachment do matter in psychotherapy—often times, in many of the same ways that John Bowlby would have predicted. According to Bowlby (1988), “unless a therapist can enable his [sic] patient to feel some measure of attachment security, therapy cannot even begin” (p. 140). Similarly, Mikulincer and Shaver (2007) argued that, in the same manner that attachment orientations affect close interpersonal relationships, attachment orientations of both clients and therapists may “affect the quality of the client-therapist relationship, determine clients’ reactions to therapists’ interventions and therapists’ reactions to clients’ disclosures, and thereby bias the therapeutic process” (p. 454). Thus, an attachment informed perspective would argue that attachment security serves as the bedrock of the therapeutic process.

For instance, attachment theory predicts that client attachment orientations would be strongly associated with therapy process and outcome (Mikulincer & Shaver, 2007). Indeed, researchers have consistently reported that client attachment security is positively related to the working alliance, whereas client attachment insecurity has been found to be detrimental to working alliance ratings (e.g., Eames & Roth, 2000; Kanninen, Salo, & Punamaki, 2000; Kivlighan, Patton, & Foote, 1998; Mallinckrodt, Gantt, & Cobble, 1995; Mallinckrodt, Porter, & Kivlighan, 2005; Parish & Eagle, 2003; Satterfield & Lyddon, 1995). 

Similarly, researchers have systematically reviewed studies that examined the relationship between client attachment orientations and therapy outcome (Berent & Obegi, 2009; Levy, Ellison, Scott, & Bernecker, 2011). For example, Levy and colleagues (2011) conducted a meta-analysis of 19 studies (n = 1,467) examining client attachment and psychotherapy outcome. Overall, they reported that clients with higher attachment anxiety scores showed the least improvement in symptoms, whereas more secure clients had more favorable outcomes. Somewhat unexpectedly, client attachment avoidance was not significantly associated with treatment outcome. Similarly, Berent and Obegi (2009) reviewed 13 attachment-outcome studies and concluded that, at “the end of psychotherapy, secure clients tend to function better, and their overall symptoms are less severe” (p. 473), although they cautioned that positive outcomes may result because secure clients enter treatment “less disturbed than their insecure counterparts” (p. 473). They further noted that there may be differential patterns of treatment response across client attachment orientation and that there was emerging evidence that fearful clients may take longer to benefit from therapy. 

Although psychotherapy researchers have an increased understanding about client attachment orientations and their influence on the process and outcome of therapy, relatively less is known about the impact of therapist attachment. As theoretically expected, initial findings generally suggest that therapist attachment insecurity interferes with the process and outcome of psychotherapy (Slade, 2016), but there are some exceptions. In a systematic review of 11 studies, Degnan, Seymour-Hyde, Harris, and Berry (2016) reported preliminary evidence that therapist attachment style has the potential to influence the working alliance and outcome, but some individual study findings were mixed. In one study, Bruck, Winston, Aderholt, and Muran (2006) found a direct relationship between therapist attachment security and improved therapy outcomes whereby higher attachment security was moderately correlated with improved client-reported interpersonal problems. In a second study, Schauenburg and colleagues (2010) found that higher therapist attachment security was associated with clients’ improved post-therapy symptoms. In a third study, Sauer, Lopez, and Gormley (2003) found that therapist attachment anxiety had a positive impact on client working alliance following the first session. Interestingly, this process, however, reversed itself across time—therapist attachment anxiety had significant negative effects on the working alliance across time. Sauer and colleagues posited that a working alliance honeymoon period quickly waned when clients were working with more anxiously attached therapists. In contrast, Ligiero and Gelso (2002) found no association between therapist attachment and working alliance. 

Despite the attraction of attachment-informed psychotherapy, the progress of our understanding is slowed by measurement-related issues. We believe that the rate of attachment research is influenced by the difficulty of accurately measuring therapist attachment orientations. For example, although the Adult Attachment Interview (AAI; George, Kaplan, & Main, 1985) is widely used and considered the gold standard for adult attachment measures, it is somewhat unwieldy as it requires extensive training and certification. This semi-structured interview yields an individual’s state of mind (or internal working models) related to attachment by assessing general and specific childhood recollections. Other attachment researchers have relied on self-report measures such Experiences in Close Relationships (ECR; Brennan, Clark, & Shaver, 1998) to target respondents’ perceptions of their general experiences in close or romantic relationships. This measure yields two primary attachment dimensions—i.e., attachment anxiety and attachment avoidance. Slade (2016) notes that while self-report instruments allow for the evaluation of dynamic change, they lack the potential for more in-depth examination of attachment-related processes that are provided by interview-based measures. While commonly used, these measures may tell us little about how therapist’s attachment actually shows up in relational behaviors during therapy sessions. 

To address this measurement gap, Talia and colleagues have recently published a groundbreaking measure of therapist attachment security called the Therapist Attunement Scales (TASc; Talia, Muzi, Lingiardi, & Taubner, 2018). This methodological innovation is used to analyze therapists’ attunement and attachment status in any given psychotherapy session. According to Talia and colleagues (2018), therapist attachment experiences activate caregiving responses in therapy. By coding transcribed therapy sessions in this way, Talia has demonstrated that it is possible to reliably determine therapist attachment style by measuring moment-by-moment discursive, relational behavior within a single psychotherapy session. This coding system yields three therapist attachment classifications: secure, preoccupied, and dismissing (avoidant). Encouragingly, the attunement scales are highly convergent with attachment scores from the AAI. Finally, as the authors predicted, the initial validation study found that therapist attachment classifications were highly associated with distinct ways that they attuned to their clients. According to Talia and colleagues, attunement is described as “how therapists communicate and reflect the patient’s internal states” (p. 3).

Under the direct guidance of Alessandro Talia, three doctoral students at WMU are currently training to become reliable coders of the TASc. Our students have noted that just being part of this training has positively impacted their clinical work. For example, they are now more mindful of how their own utterances may serve to promote or discourage attunement.

This past spring, our research team launched the first study in the United States using the TASc. We recently presented initial findings at Society for Psychotherapy Research (SPR) annual meeting in Argentina (Sauer et al., 2019). In this study, we used a naturalistic design to examine how therapist attachment is associated with working alliance and therapy outcomes in a psychology training clinic. In this study, therapy outcome was evaluated with the Outcome Questionnaire (OQ 45.2; Lambert et al., 1996) at intake and then before every subsequent session. The Working Alliance Inventory Short-Form (WAI-SF; Hatcher & Gillaspy, 2006) was used to measure working alliance prior to the second session and every subsequent session. At this point, we have 17 therapy dyads coded, consisting of six therapist trainees and 17 clients. 

Our preliminary findings revealed a number of interesting findings. First, only one of six trainees was coded as insecure (avoidant), whereas the five remaining trainees were coded as secure (no trainees were coded as preoccupied). While this is theoretically good news for clients, the lack of variation in therapist attachment security made our initial analysis somewhat thorny. We decided to collapse our dyads into two attachment groups: avoidant and secure. Two clients were seen by the one avoidant therapist and 15 clients seen by five secure therapists. We next collapsed the mean working alliance scores for the avoidant therapist and the secure therapists across the first five therapy sessions. It appears that our avoidant therapist is clearly struggling to develop an early working alliance. That is, she has lower average working alliance ratings than the secure therapists (see Figure 1). While this early finding is tentative in nature, growth modeling analysis will soon be conducted on the full sample. Consistent with Talia and associates (2018), we expect that these findings will reveal how therapist attachment-related behaviors, rather than general dispositions, affect relational and outcome aspects of clients in psychotherapy. 

Perhaps the most important implication of the future use of this revolutionary attachment measure points to its implications on training therapists. If secure therapists are indeed more effective, it will help us identify specific therapist responses that are associated with improved attachment security and attunement in psychotherapy. Armed with this knowledge, the obvious next step would to teach therapist trainees ways to improve their levels of attunement during psychotherapy sessions. Thus, the TASc, could be used to not only to track therapeutic interventions retrospectively, but also to train therapists to deliver interventions in a more secure and effective manner. 

+4

0

Eric M. Sauer, Ph.D.

Eric Sauer is a Professor of Counseling Psychology and the Director of the Center for Counseling and Psychological Services (in Grand Rapids) at Western Michigan University.