Renegotiating tasks or goals as rupture repair: A task analysis in a cognitive-behavioral therapy for personality disorder

Abstract

Objective:

The purpose of this study was to provide some definition of rupture repair in a cognitive-behavioral therapy (CBT) for personality disorders, specifically how treatment tasks or goals are renegotiated.

Method:

Following a task analysis, a rational model was developed with the support of an expert panel. An empirical analysis was conducted on six CBT cases sampled from a clinical trial that included personality disordered patients and treatment adherent therapists. Two sessions from each case indicating rupture repair were selected, based on patient and therapist ratings of the Working Alliance Inventory - 12 item version (WAI-12) and rupture presence. A qualitative analysis of the sessions was conducted with the support of the observer-based Rupture Resolution Rating System (3RS – 2022).

Results:

The empirical analysis provided some support for many of the stages defined in the rational model, but less support for the hypothesized sequences of stages. A rational-empirical synthesis yielded a revised model that suggested therapists combine various strategies in rupture repair in a variety of ways, not necessarily in consistent sequences.

Conclusions:

The renegotiation of tasks and goals in this CBT sample was variable. The importance of responsiveness and the need to validate the rational-empirical model were highlighted.

Since its transtheoretical reformulation (Bordin, 1979) and recognition as the “quintessential integrative variable” (Wolfe & Goldfried, 1988), there has been extensive research on the alliance demonstrating its predictive relationship to psychotherapy outcome, a good deal including cognitive-behavioral therapies (see Flückiger et al., 2019). Over the past 30 years, we have also witnessed a growing body of research on rupture repair as a critical transtheoretical change process (see Eubanks et al, 2019; Muran, 2019). Much of this research has been based on changes in pre- and/or post-session ratings of the alliance and their predictive relation to outcome, including mediational efforts to define rupture repair as a change mechanism. One significant limitation of this research is it provides little on the actual processes involved in rupture –“when/then” data that would provide clinicians more useful information on what to do when a rupture occurs. There have been more intensive analyses of in-session processes applying mixed methods, specifically task analyses (Benitez-Ortega & Garrido-Fernandez, 2016; Greenberg, 2007; Pascual-Leone et al, 2009) that aimed to provide such data and to define patient and therapist processes involved in rupture repair as a change event, which we will consider in this paper.

Bordin’s (1979) widely recognized definition of the alliance suggested it is comprised of purposeful collaboration (agreement on treatment tasks and goals) and the emotional bond between the patient and therapist (affective attunement and mutual respect). Safran and Muran (2000) elaborated on this definition by suggesting it allows for the recognition that patient and therapist are ongoingly negotiating their respective needs for self-definition and relatedness (or agency and communion: see Bakan, 1966; Blatt, 2008; Bieling et al., 2000) at an implicit level while negotiating the tasks/goals and their bond at the explicit level. According to these formulations, rupture has been defined as any difficulty collaborating on tasks/goals, deterioration in the bond, and breakdown in the negotiation of needs between patient and therapist. More specifically, ruptures can be distinguished between withdrawal and confrontation markers (Eubanks & Muran, 2022; Muran & Eubanks, 2020; see also Safran & Muran, 2000). Withdrawal markers are movements away from the work or the other (i.e., appeasing or isolating behaviors) that include patient or therapist shutting down the work, avoiding the work, or masking real experience: These reflect pursuits of relatedness at the expense of self-definition. Confrontation markers are movements against the work or the other (i.e., attacking or controlling behaviors) that include patient and therapist complaining/criticizing, pushing back or controlling the other: These reflect pursuits of self-definition at the expense of relatedness.

The emphasis on difficulty with collaboration on tasks or goals, rather than simply disagreeing on tasks and goals, is an important component of Safran and Muran’s conceptualization of rupture that has particular relevance for a cognitive-behavioral therapy (CBT), where the overarching principle of collaborative empiricism centers the active engagement of both patient and therapist as they “share the work” in therapy (Tee & Kazantzis, 2011). The distinction between disagreement and difficulty with collaboration is critical: A respectful disagreement can be part of effective collaboration, and a superficial, deferential agreement would not constitute true collaboration (Eubanks, 2022). Key interventions and tasks in CBT (e.g., guided discovery and Socratic method, self-monitoring and cognitive restructuring, behavioral exercises and experiments) are optimally enacted in the context of a collaborative therapeutic relationship that includes a strong alliance (Kazantzis et al., 2017; Okamoto et al., 2019). There is a growing body of research that supports the important role of the alliance in facilitating effective interventions in CBT, including evidence that a strong alliance mediates the effect of therapist competence on symptom improvement in CBT for a variety of disorders (Impala et al., 2022; Weck, Grikscheit et al., 2015).

Rupture repair has been defined by strategies to re-establish 1) collaboration on the tasks or goals of treatment or 2) emotional understanding and mutual respect between patient and therapist (Safran & Muran, 2000). Based on a review of the literature, both practical and empirical, Muran and Eubanks (2020) have specifically identified three possible pathways to rupture repair, each beginning with some acknowledgement of the rupture (either implicit or explicit) and ending with some provision of a new corrective or differential learning experience. It is important to recognize that these pathways or models are representational and in reality the repair process is complex and can involve an intersection of these (Muran, 2019; Muran et al., 2022). The first two pathways can be considered immediate strategies because they are focused on immediately responding to the rupture and taking some corrective action to get treatment “back on track.” The first involves simple clarifications or validations of feelings or intentions: often two-turn exchanges in which a therapist explains a position, corrects a misunderstanding, or supports the truth value of experience. These are responses that redress a misattunement and result in reattunement, which can challenge and correct beliefs and expectations that one cannot be understood by another. The second pathway involves the renegotiation of tasks or goals, any change in the work or in its direction, which can provide the patient with the experience of being accommodated by another, of being able to negotiate differences. The third pathway is considered expressive because it involves the exploration of the rupture, including patient and therapist respective contributions (their emotional states and behavioral actions) towards a clearer expression and recognition of implicit needs for self-definition and relatedness. This pathway was the subject of an earlier task analysis on rupture repair (described in the next section).

Expressive Model: A Review

A previous task analysis, involving a series of single-case studies, operationally defined a stage-process model and confirmed rupture repair as an expressive process (Safran & Muran, 1996). The analysis included 16 matched sessions from eight cases of an integrative cognitive therapy that included interpersonal and emotion-focused principles and strategies. Sessions were identified by post-session measures, coded with process measures of interpersonal behavior and emotional involvement, and submitted to sequential analyses. The resulting stage-process model depicts the different interventions or positions that therapists should pursue in response to various patient states or positions, including 1) addressing the rupture marker, 2) exploring the rupture experience (clarifying any mixed expressions: qualified assertions or angry expressions of hurt), 3) exploring any avoidant movement away from communicating about the rupture, and finally 4) recognizing the patient’s expression of an implicit need. The key intervention principle in this regard is to establish communication about the communication process, or metacommunication (Kiesler, 1996).

The typical progression in the withdrawal resolution process was shown to move from increasingly clearer expressions of negative sentiments to self-assertion. The typical progression in the confrontation resolution process was from expressions of anger to hurt and disappointment to vulnerability and contacting the need for nurturance. The essential tasks for the therapist to facilitate this movement is to empathize, to remain nondefensive, and to take responsibility where appropriate. Throughout such progressions, there are often shifts away, movements that reflect the patient’s anxiety about expressions of assertion or vulnerability, which the therapist should explore. It is important to note that this research suggests that resolution or productive process does not require progression through all these pathways and reaching all these patient states, especially within any given session. Instead, any exploration of a rupture or avoidance in and of itself can be experienced by patients as very meaningful (see also Safran & Muran, 2000).

There have been other task analyses that have provided some support for this expressive model. Specifically, two studies involving the integrative cognitive-analytic therapy demonstrated the value of addressing and exploring a rupture (Bennett et al, 2006; Daly et al, 2010). In two studies of cognitive-behavioral therapy, one demonstrated the value of implicit recognition of rupture and then change or revision of the task (Aspland et al, 2008); the other promoted addressing and exploring in rupture repair, then the possibility of changing or revising the task (Cash et al, 2014). Finally, two other studies, one psychodynamic in its approach, the other emotion-focused, supported the value of addressing and exploring the rupture (Agnew et al., 1994; Swank & Wittenborn, 2013; see also Hill, 2010, for a review of qualitative research that provides similar support for addressing and exploring).

Specific Aims

In this study, we sought to further our understanding of the pathways to rupture repair by conducting a task analysis (specifically the discovery phase) on the second pathway, the renegotiation of tasks or goals in a cognitive-behavioral therapy (CBT) for personality disorders. Consistent with the task analytic paradigm (see Greenberg, 2007), our application of the discovery phase began with the development of a rational model with input from a panel of CBT experts. We then conducted an empirical analysis of the rational model using session data collected from CBT cases in a clinical trial. Finally, we concluded by synthesizing a rational-empirical model. We followed an iterative process of moving back and forth between hypothesized and observed performances to build a rational-empirical model as our outcome. Our overall objective was to operationally define a stage-process model of task/goal renegotiation to complement the previous effort regarding the expressive repair process and to set the stage in a stepwise fashion typical of the task analytic paradigm for testing our model in a validation phase with another sample.

Method & Results

Rational Model

We hypothesized a rational model that depicts three possible tracks (A, B, C) that therapists can follow when trying to renegotiate therapy tasks or goals with a patient. Figure 1 illustrates the stages (therapist strategies and patient responses) in these three tracks (sequences of strategies and responses). Each track begins with an acknowledgement of some form of disagreement between patient and therapist on a treatment task or goal.

Figure 1.

A rational stage-process model regarding renegotiating tasks/goals

Track A is a straightforward, problem-solving, non-exploratory approach. It involves the therapist detecting a disagreement on a task or goal, proposing an alternative, and establishing an agreement on the alternative. For example, the therapist picks up on the patient’s reluctance to do an exposure exercise and without discussion makes an alteration to the exercise that the patient more readily agrees to.

Track B includes some exploration of the problem and possible solutions. It involves the therapist clarifying any internal (thoughts or feelings) or external (behavioral or circumstantial) obstacles for the patient in completing a task or goal, exploring an alternative, and then establishing agreement on the alternative. For example, the therapist asks the patient what got in the way of completing a thought record; the patient in turn discloses that she does not see the point in it (“I already know what I’m thinking”); then the therapist invites a collaboration to explore and devise a self-monitoring task that makes more sense to the patient. In some instances, a patient might become anxious and avoid such an exploration (perhaps for fear of the therapist’s frustration), and the therapist will have to recognize this avoidance and redirect the patient back to exploring an alternative.

In Track C, the therapist invites collaboration by sharing their reasoning with the patient. The therapist moves from clarifying negative sentiments to explaining the rationale for the task or goal before establishing agreement on the original task or goal. For example, the therapist might provide a better explanation for the therapeutic value of keeping a thought record.

As previously mentioned, Muran and Eubanks (2020) have suggested that these immediate repair tracks can be powerfully transformative (arguably equal to an expressive repair), providing the patient with a new, differential, or corrective learning experience. We constructed a panel of expert CBT scholars who reviewed and confirmed the stage-processes in this model and provided minor revisions regarding stage headings, which we incorporated into the model. The panel consisted of Amelia Aldao, Louis Castonguay, Doris Chang, Robert DeRubeis, Raymond DiGiuseppe, Marvin Goldfried, Nikolaos Kazantzis, Lata McGinn, Shelley McMain, Michelle Newman, Jacqueline Persons, and William Sanderson. They all consented to have their names revealed.

Empirical Analysis

We submitted our rational model (revised based on expert feedback) to an empirical analysis by selecting 12 sessions from an archival sample of six CBT cases (two sessions per case) that were part of a randomized controlled trial (RCT: Muran et al., 2005; Muran et al., 2009). The aim in this step was to corroborate what was hypothesized in the rational model with what could be observed in a sample purposefully selected for the task of interest –that is, renegotiation of a therapy task or goal. Following the task analytic paradigm, the empirical analysis involved a descriptive qualitative analysis conducted by two observers who worked together towards consensus (Greenberg, 2007).

Data sample

The data set used for sampling in our empirical analysis was collected from a clinical trial conducted in the 1990s at a major medical center in New York City (Muran et al, 2005; Muran et al 2009). In the RCT, 41 cases of a 30-session manualized CBT for personality disorders (Turner & Muran, 1992) were compared to cases in two other time-limited treatments. The RCT found equal efficacy across the three treatments and higher early alliance for the CBT condition (N = 128). The CBT condition applied a schema-focused model (Beck et al., 1990) that begins with the establishment of a case formulation (Persons, 1989) and includes interpersonal schemas and cognitive-interpersonal cycles (Safran & Segal, 1990) as part of the formulation. The treatment protocol then entails two intervention phases: (a) symptom reduction, in which Axis I conditions are addressed, and (b) schema change, in which core beliefs are reconsidered. Both phases include traditional cognitive–behavioral strategies, including self-monitoring, cognitive restructuring, behavioral exercises, and experimentation. The therapeutic relationship is founded on collaborative empiricism, and alliance ruptures are approached as examples of patients’ presenting problems (see Newman, 1998). In this trial, the patients presented with an Axis I anxiety or mood disorder (87%) and an Axis II personality disorder (100% Cluster C or NOS) on DSM–IV (APA, 1994) as assessed by the Structured Interview for DSM–IV-Axis I & II (SCID: First et al, 1995). The therapists in the RCT consisted of attending psychiatrists, supervising psychologists, psychiatry residents, and psychology interns.

Case and session selection

From the pool of 41 CBT cases included in the original clinical trial (Muran et al, 2005), 16 cases conducted by five therapists were selected for this study. These therapists met the standard for adherence to the CBT manual based on randomly sampled sessions (other than those studied here) from each third of the treatment protocol (see Muran et al, 2005). From these cases, 12 were identified as having rupture and repair sessions (described below). Of these 12 cases, only six had videotape availability that covered the identified sessions for empirical analysis. Six cases exceed the minimum standard of three for conducting a task analysis (Greenberg, 2007), and hence were sufficient for the purposes of this study.

Sessions were identified as “rupture” and “repair” sessions respectively based on patient and therapist ratings on a postsession questionnaire that consisted of multiple measures of session impact and therapeutic alliance (PSQ: Muran et al., 1992). More specifically, these rupture and repair sessions were identified based on the Working Alliance Inventory - 12 item version (WAI - 12: Tracey & Kokotovic, 1989) and a series of direct rupture report items (which included a rating of rupture intensity and an open-ended narrative description of the rupture: see Muran et al, 2009). Efforts to identify ruptures began with control charting of patient and therapist ratings on the WAI-12 (see Eubanks et al, 2012; Lipner et al, 2022): that is, time-series analysis that graph fluctuations in working alliance scores against the average working alliance score for the dyad across treatment. Using this method, we considered a session as marking patient- or therapist-rated “rupture” when a patient or therapist alliance rating dropped low enough to cross the lower control limit, representative of a dip of at least two standard deviations below the average alliance score for that dyad. We considered patient- or therapist-rated “repair” to be marked by a subsequent session that rose above the lower control limit and towards the average alliance. We aimed to identify rupture and repair sessions by requiring 1) convergence between patient and therapist-rated rupture and repair identified via control charting of alliance ratings or 2) confirmation of a patient or therapist-rated rupture or repair (identified by control chart) with a patient or therapist direct rupture report on the PSQ.

The final sample selected for inclusion in this study consisted of six patients and four therapists (two therapists had two patients in this sample, and two therapists each had one). In Table 1, we present the descriptive data for the six cases: specifically, basic demographics for each patient and therapist, patient primary diagnoses on Axis I and II, therapist professional experience (postdoctoral years) and case number conducting the CBT treatment protocol in this study, and various treatment variables, including sessions completed, sessions identified for coding rupture and repair, and treatment outcome, which is determined by repeated measurement on the Symptom Checklist - 90 Revised (Global Severity Index: Derogatis, 1983) at intake and termination, calculating Reliable Change (Jacobson & Truax, 1991; Muran et al, 2005). All four therapists identified CBT as their primary orientation.

Table 1.

Case descriptives

Case ID	Patient Demographics	Patient Diagnosis	Therapist Demographics	Therapist Experience	Treatment Variables
461	45 year old White male	Mood disorder Obsessive-compulsive PD	40 year old White male	PhD 13 yrs experience 2nd case in study	30 sessions completed Sessions 8-9 coded RCI on SCL: −3.72*
494	33 year old White male	V-code Personality disorder NOS	42 year old White male	PhD 15 yrs experience, 3rd case in study	28 sessions completed Sessions 6-7 coded RCI on SCL: N/A
503	28 year old, Black female	Mood disorder Personality disorder NOS	35 year old White female	PhD 4 yrs experience, 2nd case in study	30 sessions completed Sessions 11-12 coded RCI on SCL: −5.06*
573	31 year old, Asian female	Mood disorder Personality disorder NOS	33 year old White female	PhD 3 yrs experience 2nd case in study	25 sessions completed Sessions 18-19 coded RCI on SCL: N/A
604	33 year old White male	Anxiety disorder Avoidant PD	30 year old White female	MA predoctoral intern, 2nd case in study	30 sessions completed Sessions 11-12 coded RCI on SCL: −2.33*
696	29 year old White female	Mood disorder Personality disorder NOS	36 year old White female	PhD 6 yrs experience, 7th case in study	30 sessions completed Sessions 22-23 coded RCI on SCL: −6.30*

Note: Only primary diagnoses on Axis I and Axis II are reported. Cases 461 and 494 were conducted by the same therapist; Cases 573 and 696 were conducted by the same therapist. For therapist experience, degree, post-doctoral years, and case number in the trial are reported. Sessions coded for the rupture repair process with the 3RS were indicated along with sessions completed. Reliable change index (RCI) was calculated on SCL-90R (GSI) intake and termination ratings,

^*p < .05.

In-session observation

A descriptive (qualitative) analysis was conducted on the 12 sessions from the six cases by two coders, the third and fourth authors, who were doctoral-level clinical psychologists with primary integrative psychotherapy orientations (including training in CBT and dynamic therapies). Coders were provided with the rational model and asked to look for evidence of ruptures related to treatment tasks/goals and to observe therapists' responses to these ruptures in the selected sessions: more specifically, to confirm or refine the rational model, based on an analysis of possible versus actual behaviors. Coders were previously trained on the Rupture Resolution Rating Scale (3RS: Eubanks, Muran & Safran, 2015) and used it as a support tool to inform their observations of rupture markers and repair processes. The 3RS is an observer-based measure of rupture markers (specifically withdrawal and confrontation, see Table 1 for examples) and repair strategies (including changing tasks or exploring ruptures) that has been shown to be applied with high inter-rater reliability and to predict premature termination to treatment (Eubanks et al, 2019). Most recently, it was revised to additionally assess therapist contribution to rupture and patient contribution to repair, as well as to track if the patient and therapist are "working together" in a collaborative (engaged and respectful) way (Eubanks & Muran, 2022). In this study, the coders reviewed their codes together and discussed their findings and worked toward consensus. For example, they drew on the 3RS confrontation marker of complaint or criticism about treatment tasks to help them identify disagreements about therapy tasks or goals, and then drew on 3RS repair strategies (e.g., changing the task; providing a rationale for a task) to help them identify and describe therapists' responses to these disagreements. They also aimed to identify any repair processes they observed that were not included in the rational model. All critical processes were transcribed by the coders and carefully considered and discussed by all the authors after the coding was complete. All authors contributed to the final description and presentation of the results of the rational analysis.

Results

Here we present the empirical evidence from our cases regarding the various stages and tracks hypothesized in our rational model of renegotiating tasks or goals. We note in parentheses the number of cases in which each stage was observed. With the extended examples, we also note markers of collaboration (working together), rupture, and repair by indicating relevant 3RS codes.

Hypothesized stages.

We begin with the stages in our model, with the various interactions between patient and therapist with each stage:

Acknowledgement of disagreement.

All the cases (6) evidenced some rupture in the form of disagreement on the tasks/goals of treatment (e.g., topics, exercises, and experiments) and then various strategies to repair (as we will describe below). Acknowledgement of disagreement by the therapists was primarily implied (4) by their efforts to repair (which in all six cases included as a first step some repeat effort to pursue and establish agreement on the original task). There was one instance in which the patient announced at the start of the session, “I was feeling a little apprehensive about coming in this morning” (Case 461). There were also a few instances (in Case 696) in which the therapist explicitly acknowledged rupture through observations and self-disclosures, such as: “I feel like right now we’re having a hard time” and “I feel lost” or “I’m still confused.” These expressions seemed to come at a point when all other repair strategies failed to establish an agreement on a task. The ruptures observed across the six cases included both withdrawal markers, such as patient shutting down the work (minimal response or avoidant denial of a feeling state manifestly evident), patient and therapist avoiding the work (abstract communication and topic shift), and patient masking real experience (deference/begrudging compliance to a therapist direction and content/affect split), and confrontation markers, such as patient pushes back (rejects therapist’s ideas and defends self: “I don’t want to talk about this” and “I don’t want to dwell on that” - Case 573; “I thought I was saying it” - Case 503), therapist controls patient (overly directive and pushes ideas on patient: non- collaborative efforts that go beyond challenges typical of cognitive psychotherapies).

Proposal of alternative.

Half of the cases (3) included explicit examples of the therapist proposing an alternative task. To illustrate, in one instance a patient expressed difficulty with a relaxation exercise, “It slows me down a bit… but it’s difficult for me to control my thoughts,” to which the therapist proposed, “Okay, let’s talk about that,” and directed the patient away from the exercise and to a discussion of the patient’s difficulty with his thoughts (Case 604). Later during the assignment, the patient proposed an alternative, and the therapist agreed, “Maybe I’m giving you the wrong thing, we need to tailor it to you.” Although the other half of the cases did not explicitly evidence such a strategy, they did have examples of the therapist making minor, more subtle “refinements” to the original task.

Agreement on alternative.

All the cases (6) evidenced examples of the patient eventually agreeing with a “refined” or “alternate” task as defined above. In most of the cases (5), these movements towards agreement were relatively subtle and involved an elongated process. In one case (Case 604), there were more obvious and succinct examples, marked by such therapist expressions as: “I’m giving you too big a thing… it may not be you… I want it to be realistic [about your homework]…” Or in another instance in the same case: “It may be helpful to talk about [an alternate task]… You seem to be having trouble with [the original task].” Both of these expressions (which were further examples of proposing an alternative) were followed by the patient earnestly agreeing and engaging with the therapist’s proposal.

Clarification of obstacles.

Half the cases (3) included an effort to clarify obstacles to engaging in a particular task –in all instances internal obstacles. In one case (Case 503), the therapist invited the patient to explore her difficulty describing feelings about an issue: “What do you make of this being so hard?” In another case (Case 573), the therapist inquired about the patient’s difficulty with attendance as she no showed for the previous session, to which the patient revealed her experience of therapy as “a chore.” In yet another (Case 604), the therapist checked on the patient’s understanding of a task, to which the patient expressed some “doubts” and the therapist followed in turn with an invitation to explore, “That makes you feel [what]?”

Exploration of alternative.

Half the cases (3) evidenced an instance of exploring an alternative task. In one case (Case 573), this was marked by a therapist's suggestion, “Okay, let’s not continue talking about that. What do you want to talk about?” In another case (Case 604), the following exchange illustrated this approach to rupture repair in response to the patient expressing some doubt about an exercise (specifically, considering the veracity of a cognition):

Therapist: That’s an important piece to bring up! Maybe it’s not you, maybe I didn’t give you the right exercise to do, and we should have done it differently. Now let’s not get into what we should have done… What is right for you? (Therapist repair markers: Therapist validates patient’s experience of the rupture; Therapist acknowledges contribution to rupture; Therapist proposes/discusses changing task/goal. Working together marker: Therapist validates patient; Therapist curious/engaged.)

Patient: Maybe I could do an assessment of what happened during the day at the end of the day, so I have a more global view. (Patient repair marker: Patient discusses changing task/goal. Working together marker: Patient engaged.)

Therapist: Ok let’s do it this way. It’s more you. (Therapist repair marker: Therapist changes task/goal. Working together marker: Patient and therapist collaborating on work of therapy.)

Redirection of avoidance.

There were no instances in this data set in which the therapist redirected the patient who was avoiding some aspect of the repair process back to the task of therapy.

Explanation of rationale.

Half the cases (3) included an example of the therapist explaining the rationale for a task. In one case (Case 604), this was illustrated by the therapist stopping to clarify as follows: “I wanna bring things back to how automatic thoughts influence behavior…” Other examples are forms of psychoeducation where the therapist tries to explain how a patient “distances” herself from her feelings or how certain cognitions may be “distortions” of what may be happening in reality or what may be seen by others (Case 503).

Agreement on original task.

There were several cases (4) in which agreement on the original task was established or re-established, and these appeared subsequent to the therapist explaining the rationale for the task and persisting with the task (Cases 461, 494, 573, 604).

Hypothesized tracks.

Here we turn to the tracks in our model, to address to what extent the stages moved according to the three hypothesized tracks. The prevailing evidence indicated that therapists applied the strategies described above in more varied ways than hypothesized in the rational model. Tracks A, B, C were evident only in parts. Instead, we sometimes saw Acknowledgement of disagreement lead directly to Agreement on alternative, sometimes to a Proposal of alternative, sometimes to Exploration of alternative, sometimes to Clarification of obstacles. We saw Proposal of alternative, Exploration of alternative, Clarification of obstacles at various times lead to Agreement on alternative. We also saw instances of Explanation of rationale leading to Agreement on original. In short, we saw various combinations of repair strategies (defined in the rational model) as therapists attempted to respond to rupture (or lack of agreement in tasks/goals). We also saw a good deal of back-and-forth between implicit recognition of rupture and implementation of a repair strategy as therapists tried to adapt and tailor their efforts.

Other observed processes.

In addition, we observed a couple of noteworthy strategic processes not defined in the rational model of renegotiation. First, we observed that therapists often tried more than once to pursue a task with their patients or push their patients a bit before resorting to any of the strategies defined in the rational model. Here is an example where the therapist confronts the patient about her report to push a point about the patient’s difficulty expressing anger (Case 503):

Therapist: I’m not hearing you, at least in the way you’ve described so far, I’m not hearing you say to your sister that you were frustrated by what had taken place in exactly those kinds of words or that you were angry. (Therapist challenges patient.)

Patient: I thought about saying it over the phone (patient giggles), but I said it when she got home (giggles more). I might actually cause I was like, you know, okay, fine, but I was all upset (adds something unintelligible). [Patient rupture markers: Patient masks her real experience (content/affect split); Patient pushes back (defends self).]

Therapist: Did anyone know, see it? (Therapist persists with line of questioning.)

Patient: Oh, they knew (giggles). My parents, my dad, he absorbs all. [Patient rupture markers: Patient masks her real experience (content/affect split); Patient pushes back (defends self).]

Therapist: How did he know you were upset? (Therapist persists and risks rupture: Therapist controls/pressures patient.)

Patient: Cause I was “fussing” (giggles more). When I was on the phone, I was fussing. I don’t know if I said it out loud. [Patient rupture markers: Patient masks her real experience (content/affect split); Patient pushes back (defends self).]

Therapist: I think in terms of our work here, in terms of dealing with anger better, one of the concerns I have is that when you’re feeling angry, what you’re calling “fussing,” because you have this belief that you don’t wanna upset everybody else, that you try not to rock the boat, in this situation, as in the past, what might have happened in this situation, like in the past, is that you fuss, and it seems very exaggerated to you when you’re doing it because you’re not used to it. But my concern is that it’s very subtle and that other people observing you, or other people that are there with you may not be understanding how you’re feeling or that you’re experiencing anger… Is that a possibility? (Therapist repair marker: Therapist explains rationale.)

Patient: They might not, yeah, I guess, they might not recognize it as anger but see it as being contrary or what have you… (Working together marker: Patient and therapist collaborating on work of therapy.)

In this example, the therapist persisted with a line of questioning that could have resulted in rupture, but instead her effort landed well as the patient ultimately began working with her on the formulation.

Second, we observed many therapist efforts to modify or refine a task (sometimes in their persistence of a particular task as described above). Such refinements often involved breaking down the task into smaller and easier tasks, formulating the task differently, or making more explicit suggestions instead of posing questions. In one case (494), for example, the therapist challenged a patient’s interpretation of her boyfriend's comment about her weight, “That’s the way you think he’s judging you… What is it in what he said that made you hear that?” The patient resisted this approach (Patient rupture marker: pushes back on the therapist’s approach), prompting the therapist to revise the challenge, "Let's think about it in a different way. Let's say you date, he accepts you, you lose weight, and he says that you look better. Would that be a problem?" The question gave the patient pause, and then she began working with the possibility that she might be misattributing thoughts to her boyfriend. In another case (696), the therapist struggled to understand whether the patient's belief that she may lose her job was realistic or some distortion. The patient resisted this approach (Patient rupture markers: avoids the work and pushes back), and the therapist eventually shifted to a problem-solving approach based on the assumption that the patient may indeed lose her job, “So what are your options [if you do lose your job]?" The patient's response focused on her expectation that she would quit her job, prompting the therapist to refine the problem-solving task by asking "what would be the best way to do this?" As the patient's responses suggested that she did not understand the task, the therapist further refined it by making an explicit suggestion, "I guess I was thinking it would be better to have another job before you quit," to which the patient agreed, “I will definitely, I mean I will look for another job before I quit…” and then they began collaborating on the details of such a plan. Such refinements were common across all the cases.

Rational-Empirical Synthesis

The empirical analysis of our rational model led to two types of revisions: revising the stages and revising the tracks. For both of these revisions, we also made pragmatic changes (i.e., rephrasing most repair stages as therapist actions, building more flexibility into the model) with the aim of creating a rational-empirical synthesis that would be maximally useful for therapists.

Revised Stages

As we found that the initial Acknowledgement of disagreement was often implicit rather than explicit, and in four cases was not so much a clear or pronounced disagreement about a task, but more a difficulty being “on the same page” about a task, we renamed this stage Difficulties in collaboration on task/goal. To capture the fact that sometimes therapists did communicate explicitly about these difficulties in collaboration in order to address them, we added a stage of Explicitly acknowledge difficulties in collaboration.

The rational model included two stages related to introducing an alternative task or goal: Proposal of alternative task/goal and Exploration of alternative task/goal. In our sample of cases, we did not observe a clear-cut distinction between these two processes: the cases in which an alternative was proposed, it was also explored. We concluded that what we were observing was better captured by Discuss alternative task/goal (inclusive of proposing and exploring). We also observed that in four out of six cases, therapists introduced subtle modifications to the task that were not announced in any explicit way. To capture these subtle refinements of the task, as well as times when therapists changed a task in an effort to reach agreement with the patient, we added the strategy Change/modify task/goal.

The rational model included two stages that represented a successful repair: Agreement on alternative task/goal and Agreement on original or alternative task/goal. This was another place where there was not a clear-cut distinction between these stages in our sample. Particularly given the common therapist strategy of slightly modifying the task, the distinction between the original or the alternative task was sometimes very minor. For the sake of parsimony, we combined these into one strategy of Collaboration on original or alternative task/goal.

The empirical analysis revealed an additional strategy that we had not included in our rational model: Repeat approach. In several cases, therapists’ initial response to a difficulty with a task was to “give it the old college try” and try the task again. We observed that in several cases, therapists tried this strategy multiple times, in an unsuccessful manner that the coders aptly described as “headbanging,” consistent with a therapist confrontation rupture marker of trying to control or pressure the patient. We observed that repeating the approach to the task one time was sometimes successful; however, employing Repeat approach multiple times with no success would be a failure of appropriate responsiveness (see Stiles & Horvath, 2017).

The stages of Clarification of obstacles, Explanation of rationale, and Redirection of avoidance were retained in the rational-empirical model, but rephrased as therapist actions: Clarify obstacles, Explain rationale, and Redirect avoidance.

Revised Tracks

As described previously, the empirical analysis yielded some support for our hypothesized stages but much less support for our hypothesized tracks as therapists drew from the various strategies in different ways, sometimes shifting back and forth between several strategies. Hence, we decided to revise the model (see Figure 2) to reflect that therapists encountering difficulties in collaboration on a task or goal have multiple options, and those options all have the potential to facilitate collaboration between patient and therapist. Simsilar to the rational model, we have positioned Redirect avoidance as a stage that may emerge in the context of other stages whenever patients (and possibly also therapists) are tempted to move away from a difficult conversation.

Figure 2.

A rational-empirical stage-process model regarding renegotiating tasks/goals

Once we revised the stages, we were able to identify a few sequences that were more common: three of the four therapists employed Repeat approach or Change/modify task/goal as initial steps, and then if those were not fully successful, they employed other strategies such as Explain rationale. However, given the small size of our sample, and the fact that these sequences were not illustrations of progressing through stages to reach repair, but rather examples of trying one thing, not succeeding, and then trying something else, we think it is premature to propose specific sequences as optimal paths to rupture resolution. Future research should explore and compare different sequences, as well as explore whether different types of therapy tasks or collaboration difficulties marked by more withdrawal vs more confrontation warrant different sequences of repair strategies.

Consistent with the literature on responsiveness (e.g., Watson & Wiseman, 2021), we think it is most useful for therapists to be aware of a range of possible strategies that they can draw on, based on what they know about the patient they are working with in the moment. Until we have more research on renegotiation of tasks and goals to guide us, therapists can draw on the research on principles of change to inform their selection of repair strategies. For example, therapists can adapt to the patient’s level of reactance by using less directive strategies (e.g., Discuss alternative task/goal or Change/modify task/goal) rather than more directive strategies (e.g., Repeat approach) with highly reactant patients (see Beutler, Edwards, & Someah, 2018).

Intersecting Pathways

The empirical analysis also yielded evidence of the other two pathways to rupture repair, the reattunement pathway and the expressive pathway (see Figure 1) and demonstrated how these three pathways may intersect in clinical work. As an example of the potential role of the reattunement pathway when renegotiating a task, consider Cases 503 and 494, in which the therapist challenged the patient’s interpretation of an interaction with loved ones. In Case 503, this challenge came in the context of an attuned therapeutic relationship and a strong bond and the members of the dyad were able to work together productively despite some disagreements. By contrast, in Case 494, the therapist provided little verbal empathy or validation to the patient despite her apparent distress and hurt. He focused on challenging the patient’s interpretations of her loved one’s comments in a “dry” and technical way that the patient may have experienced as invalidating. The therapist’s approach may have contributed to the patient’s response: she frequently disagreed with the therapist, appeared annoyed with him, and shifted the topic in an avoidant manner that hindered the dyad’s ability to work together.

Case 503 also provides an example of the intersection of the expressive pathway with renegotiation of task: While explaining the benefits of the patient asserting herself more with her loved ones, the therapist observed the patient becoming tearful. The patient offered a vague, intellectualized explanation for her tears (Patient rupture marker: patient avoids (abstract communication). The therapist began to explore this rupture by metacommunicating that perhaps the patient felt criticized by the therapist: “Did you think I was saying you were bad?… Did it feel critical?” (Therapist repair mark Fer: Therapist invites patient to explore the rupture). The patient said she did not feel criticized and apologized for becoming emotional (Patient rupture marker: Patient masks her real experience (deferential/appeasing). The therapist then disclosed her reaction to the patient’s tears and reframed them as a sign of progress toward the goal of asserting and expressing herself more with others: “I think it’s wonderful that you’re able to cry here because you do tend to try to cut off your feelings, that’s been one of your patterns. And the fact that you’re actually able to begin to open up and share that with me is great!” (Therapist repair markers: Therapist discloses internal experience of the rupture, Therapist links rupture to interpersonal patterns; Working together marker: Therapist accepts/validates patient).

Finally, the empirical analysis was also a good reminder that efforts to repair can cause– or contain –additional ruptures. As noted previously, we observed examples in which the strategy of Repeat approach, if employed in a rigid or nonresponsive manner, became indistinguishable from a therapist confrontation rupture of controlling or pressuring a patient, and appeared to elicit more confrontation or withdrawal from the patient. We also observed examples in which it was unclear whether the patient and therapist were achieving repair by reaching agreement on an alternative task or were colluding in a withdrawal by shifting the topic. Relatedly, the coders observed that in half the cases, in the session following the “rupture” session, the patient and therapist did not return to the topics or issues that were the source of difficulty in the prior session. Such movements are difficult to definitively define without more data (including quantitative coding of behavior turns and qualitative phenomenological interviews to provide a more granular analysis): Are these collusive avoidances, are they necessary avoidances, or are they just agreements to address other relevant issues?

Discussion and Conclusion

With this discovery phase of a task analysis, we have hypothesized a rational model (with the support of an expert panel) that defined various stages and tracks in the renegotiation of tasks/goals in a CBT for personality disorders. We consider this an important pathway in rupture repair, along with the reattunement and expressive pathways, all critical change processes across multiple types of therapy. We conducted an empirical analysis of our rational model of renegotiation by selecting sessions identified via patient and therapist self-report as involving rupture and repair processes and then conducting, intensive observations drawing on 3RS markers to provide qualitative descriptions. Our analysis provided some support for many of our hypothesized stages, but less support for the tracks in our rational model. From these findings, we conducted a rational-empirical synthesis that identified a variety of specific repair strategies that CBT therapists can employ in a variety of combinations to redress difficulties in collaboration. Our modest aim was to provide some definition for further research to validate or elaborate.

Following the task analytic paradigm (Greenberg, 2007), our study involved an iterative process of tacking back and forth between possible and actual performance and from the local to the global and back again (see also Geertz, 1983, for his qualitative method). Our approach in this study concentrated on the discovery phase to build a rational model and then conduct an empirical analysis and rational-empirical synthesis. The task analytic paradigm was introduced as a mixed methodology to define change events and to contribute to considerations of sequences (and the question of causality) in change processes (see Greenberg, 2007; Rice & Greenberg, 1984). Here we have applied the paradigm within the framework of a clinical trial, where an in-session event (or change process) is identified by intermediate (postsession-level) outcome assessment, which has been linked to ultimate (pre-to-post-treatment) outcome assessment (see Lipner et al., 2022; Muran et al., 1995; Muran et al., 2009; see also Muran, 2002, and Muran, 2019). Future efforts should continue this work towards a validation process and further consideration of what causes what, including sequential analysis of patient and therapist contributions to rupture and repair. Of course, when considering the possible transferability of our findings, our effort should be considered with regard to all our participants, including the authors and expert panelists, as well as the patients and therapists in our empirical study sample, plus the small size and the treatment parameters. A larger and more diverse sample of patients and therapists, the use of more sessions, the application of other methods and criteria for identifying rupture repair sessions or events, and the use of other third-party observers and other observer-based coding systems would increase confidence in our findings.

We would like to conclude with some considerations raised by our expert panel that had relevance for what we observed in our empirical analysis. First, three panelists (Goldfried, Kazantzis, and Persons) especially emphasized the importance of case formulation in the careful planning for intervention, which should include identifying potential obstacles to treatment, such as ruptures (see Okamoto & Kazantzis, 2021; Impala et al., 2022).

Second, two panelists (DeRubeis and Persons) emphasized the value of clearly explaining the rationale for a task, including using illustrations to this effect (e.g., doing a thought record together in-session). To the extent that our therapists took pains to explain or provide some psychoeducation, this proved effective in promoting collaboration.

Third, two panelists (Aldao and DeRubeis) raised an important question about the determination of pushing versus backing off. This has relevance to other treatment orientations, but certainly in CBT therapist challenges (with questions and exercises) are hallmark interventions. These can be characterized as confrontation, and they do require careful application. Otherwise, they become examples of therapists provoking rupture through coercion (see Castonguay et al, 1996). In our data set, there were instances in which therapists were effectively pushing their patients, and then there were instances in which they went too far and there was rupture or a break in collaboration.

This brings us to a fourth and final observation from our panel, three panelists (Chang, DiGiuseppe, and Kazantzis) highlighted the importance of therapist responsiveness, which we previously mentioned with regard to our data analysis: By this, we refer to the need to be attentive to patient abilities, emotions and ruptures as they emerge in the moment. This should be integral to CBT where there is such an explicit emphasis on collaboration (e.g., collaborative empiricism, shared decision-making, and soliciting patient feedback: Kazantzis et al, 2017; see also Muran et al, 2018, for a study of rupture-repair training in CBT). Future research should also explore responsive rupture repair in the context of patient and therapist cultural identities and power differences. (Considerations of treatment adherence and therapist flexibility are also relevant to discussions of responsiveness: see Barber, 2009, and Sanderson, 2006, for example.)

As can be seen from the comments from our expert panelists, the rational-empirical model that emerged in our study is consistent with best practices in CBT in its focus on recognizing and addressing difficulties with a treatment task in a collaborative way (Eubanks, 2022). Therapists who are guided by collaborative empiricism may avoid some ruptures around tasks and goals by effectively collaborating with the patient to select goals, facilitate engagement with tasks, and link goals and tasks effectively to a strong case conceptualization grounded in the patient’s experience. At the same time, given the fallibility of therapists, the complexity of interpersonal interactions, and evidence that ruptures are more common in CBT with patients with personality disorder diagnoses (Coutinho et al., 2014), CBT therapists should be prepared for ruptures to occur, particularly when patients have negative beliefs about others that lead them to expect criticism or invalidation from the therapist (Beck, 2005). By attending closely to the strength of their collaboration and bond with the patient, therapists will be in an optimal position to recognize ruptures and work toward repair.

Finally, we would like to consider the relationship among our three pathways to rupture repair. Our first reattunement pathway aimed to capture the micro movements from misattunement (misunderstanding) to reattunement (understanding); It is very much in line with what has been described in parent-infant research where rupture repair is invoked and the “messiness” of human interaction is dramatically enacted (see Tronick, 2007). Reattunement can be considered the glue of alliance building and rupture repair, and it was very much present in our empirical analysis. Our third expressive pathway aimed to capture the potential of using the rupture to explore implicit beliefs and underlying needs (n.b., a number of our experts –McGinn, McMain, and Newman --particularly acknowledged its value). We only saw minor expressions of it in our data set. However, what we want to clarify here on a closing note is that defining these pathways discretely should not mean that they should operate discretely or be pursued independently. Rupture repair should be understood as messy or complicated (see Eubanks et al., 2022; Muran & Eubanks, 2020) and can be achieved with various combinations of these pathways, something that we suggested up front and was evident in what we saw our therapists in this study do within our second renegotiation pathway.