An Empirical Perspective on the Transference in Psychotherapy: Commentary on Kernberg

Transference Work as a Change Mechanism

As in psychoanalysis, the efficacy of transference-based interventions has been a topic of debate in psychotherapy research. Some of the earliest research in this area was Kernberg and colleagues’ work on the Menninger study (Kernberg et al. 1972), which suggested that patients with significant personality pathology benefit more from expressive psychotherapy than from supportive psychotherapy or psychoanalysis. However, for a period the psychotherapy literature began to push in the other direction, suggesting that patients with impaired object relations cannot tolerate more expressive transference-based work, and that perhaps transference interpretations are better used with healthier patients (Ogrodniczuk et al. 1999; Piper et al. 1991) or in the context of a strong working alliance (Winston, McCullough, and Laikin 1993). This was bolstered by clinical theory that saw more risks than potential rewards resulting from a more interpretive therapeutic approach (for a review, see Levy and Scala 2014).

More recent work by Per Høglend and colleagues has challenged these commonly held assumptions. In the First Experimental Study of Transference Work (FEST), patients were randomly assigned to psychodynamic treatments dismantled to either include or not include transference interpretations. The researchers found that whereas patients with healthier object relations did not differ in terms of benefit in treatment as a function of transference work, patients with impaired object relations benefited from treatment that specifically included transference interpretations (Høglend et al. 2006); moreover, these effects were sustained over the long term (Høglend et al. 2008). Importantly, it was found that working within the transference was most effective for patients with impaired object relations in the context of a poor working alliance (Høglend et al. 2011), and the mechanism for change in this group was demonstrated to occur through increased insight (Johansson et al. 2010). It is of note that no patients dropped out of the treatment condition that included transference interpretations.

While not specific to TFP, these findings lend strong support to Kernberg’s model, with the level of transference work scaled to the personality organization of the patient. Contrary to clinical assumptions, this research supports Kernberg’s assertion that interpretation may be most effective in the context of an alliance that is strained by distortions in perceptions of the therapist, and that insight gained through such transference work leads to sustained change.

Of note, in the FEST study what was considered a “moderate” level of transference work included fewer interventions than might be expected, with an average of four to six instances of transference use per session. While to date TFP has not been evaluated in terms of frequency of transference interpretations, Kernberg’s clinical descriptions in the present essay suggest a much more active interpretive approach. Early research on transference use may suggest some risks associated with this level of therapist activity (see, e.g., Høglend and Gabbard 2012), with earlier correlational studies suggesting diminishing returns at the very highest levels of interpretation use (Ogrodniczuk et al. 1999; Piper et al. 1991). These findings make intuitive clinical sense, likely reflecting a process in which the therapist barrages the patient with a series of rejected interpretations. By contrast, a lower intervention “count” may reflect use of interpretations that the patient can make use of, as evidenced by the opening up of insights and new material that make further interpretations unnecessary. In fact, McCullough and colleagues (1991) found in brief dynamic therapy that interpretations that were met with affective elaboration by the patient predicted positive treatment outcome, whereas those that were met with defensiveness predicted poor outcome. Thus, while research cautions against the high quantity of transference interpretations suggested by Kernberg’s approach, ultimately it is the quality of interpretations that is most likely to predict their effectiveness.

It is important to note that what in the FEST study constitutes “transference work” would from a TFP perspective fall under the rubric of clarification and confrontation, not just interpretation (Høglend 2014). In TFP interpretations are preceded by repeated efforts to clarify the subjective experience of the patient, and to confront disparities between the patient’s verbal and nonverbal communications. Further, in TFP interpretations seek to go beyond identifying parallel dynamics in the patient’s external relationships and in-session relatedness with the therapist. The goal of interpretation in TFP is to help patients appreciate that they occupy both sides of object relations dyads that are not integrated; for instance, patients can be quite aggressive in expressing their sense of victimization, or can be quite dependent on people they treat dismissively. Further, the goal is to help the patient appreciate that one of these dyads may defend against the other; for example, a patient might rather experience the anger of feeling victimized than the terror of loss associated with feeling dependent. Interpreting the total experience of the patient’s object relations fosters integration, as the patient comes to identify with both sides of the dyad and their associated affects—experiencing the self as both victim and aggressor, as both dismissive and dependent. Research supports this assertion, as TFP has been found to lead not only to symptom change, but also to gains in level of personality organization (Doering et al. 2010). Further, reflective awareness is thought to grow as patients become aware of the total experience of their internal representations, coupled with a decreasing need to project unacceptable aspects of these dyads onto others. Research also supports this view, as changes in reflective functioning have been uniquely observed in TFP (Fischer-Kern et al. 2015; Levy et al. 2006).

An example of the different conceptualizations of transference work is found in a transcript of a late-treatment session from the FEST study; Ulberg and colleagues (2014) report the following dialogue, with their codes for the type of transference work used (see Table 5):

Of note, as the therapist begins to clearly identify the negative transference (therapist as aloof, patient as ignored), the patient retreats from this and says her concern is dissipating. While it might be argued that the therapist is making a bid toward integration of positive and negative feelings toward the therapist (i.e., simultaneously holding times she both does and does not experience the therapist as aloof), it is notable that the level of transference intervention decreases during the dialogue (rather than increase to a Category 4 interpretive link between the experience of the therapist and a dynamic conflict, or a Category 5 interpretive link between interpersonal/genetic patterns and transactions with the therapist). While in the FEST study this would nonetheless constitute transference work, from a TFP perspective this would be viewed as colluding around evading the negative transference. In fact, the therapist is almost directive in encouraging her to shut down the negative transference; while in the short term this may help her to not avoid sessions, in the long term this may perpetuate the patient’s fear that her anger (at being ignored, at feeling not cared for) is intolerable to others and must remain hidden. Thus, despite being at a relatively late stage of treatment, her aggression toward the therapist can still be expressed only obliquely (“Yes, well, we’ll see about that”). Finally, by leaning into the negative transference a TFP therapist would seek to identify the ways in which the patient occupies both sides of this dyad. For example, the patient could be seen as aloof in her deflecting observations of her feelings that “became all right kind of”; in doing so the therapist could feel ignored, perhaps leading to an enactment in which he asserts “you should remember” a more benign feeling about treatment. Thus, integration comes not from holding positive and negative images of the therapist simultaneously, but from the patient’s integrating the total experience of her ignored/aloof self.

Transference Work in Psychotherapy Process

The literature reviewed on the role of transference interpretations suggests that research needs to move beyond the quantity of interpretive behavior and instead focus on the quality of transference interpretations as responsive to the object relations dyad observed in the moment-to-moment process. Unfortunately, such empirical efforts are quite labor-intensive, and most psychotherapy process research has stayed at the level of the therapeutic alliance (Norcross 2011). While there is substantial research evaluating therapists’ general responsiveness to patients (and patients’ general responsiveness to therapists), Kernberg articulates a more specific model of what constitutes responsiveness: namely, whether the intervention corresponds to the specific organization and type of the patient’s pathology. In one of the few studies of its kind, Crits-Christoph, Cooper, and Luborsky (1988) demonstrated that use of transference interpretations that strongly converged with the patient’s core conflictual relational theme (CCRT) positively related to treatment gains. This empirical approach is consistent with Kernberg’s focus on “interpretation of both defensive and impulsive internalized object relations as they are reflected in the patient’s pathological interactions with significant others, particularly in the transference” (p. 956).

Unfortunately, in clinical trials any evaluation of the quality of treatment is often limited to assurances that clinicians have met a basic level of adherence and competence. Further, psychotherapy process research has questioned whether therapists are always doing the therapy they think they are. Ablon, Levy, and Katzenstein (2006) found that in self-identified psychodynamic therapists, many of their interventions were characterized by an independent observer to better fit a CBT prototype. Of note, this lack of adherence to a prototypical treatment may not itself be a bad thing; in fact, psychodynamic process elements were found to predict positive change in CBT treatments (Ablon and Jones 1998). Jones and Pulos (1993) developed a “smuggling hypothesis” in which they reasoned that the most efficacious psychodynamic clinicians may flexibly dip into divergent ways of working with clients, including techniques that do not conform to a standard psychodynamic approach.

Such responsiveness is one of the most distinguishing features of Kernberg’s model for working across diverse personality pathologies; despite each of these clinical interventions ostensibly representing TFP in action, the descriptions of interventions among the patients vary widely and demonstrate the need for significant flexibility in the face of distinctive object relations presentations. While the same basic techniques underlie clinical decisions, each instance necessitates manifestly different behaviors in order to elaborate the object role ascribed by the patient—be it the object of the patient’s envy, dependency, rage, or sexual longing. This type of therapist responsiveness is consistent with the research of Stiles (2009), who has demonstrated that what clinicians actually do relates as much to the interpersonal characteristics of the patient as the brand of therapy being practiced (Hardy et al. 1998). Kernberg reminds us that treatments should be grounded in principles that guide technique, but should also be responsive to the diverse object roles that different pathological presentations invite.

This is not to say that TFP lacks coherence as a treatment. In a recent study Kivity and colleagues (2019) evaluated videotaped sessions of a clinical trial of TFP; not only were the therapists rated to be moderately to highly adherent to the TFP prototype, but this treatment was also distinctive from a supportive psychodynamic therapy (as well as dialectical behavior therapy). Perhaps most important, adherence to the TFP prototype predicted growth in reflective functioning during treatment.

However, one of the great limiting factors in our understanding these change processes has been the overwhelming emphasis in psychotherapy research on borderline personality disorder (BPD), despite Kernberg’s clear articulation of the distinctive clinical presentations in cases of narcissistic and schizoid personality, each of which necessitates greater empirical focus. While narcissism is of particular clinical interest in psychoanalysis, there is shockingly little psychotherapy research addressing the treatment of patients with narcissistic personality disorder (NPD). To date there have been no randomized clinical trials of a manualized treatment for NPD. While there are now dozens of RCTs that have evaluated treatments for BPD (for a review, see Cristea et al. 2017), with some evaluating the impact of comorbid narcissistic (Diamond et al. 2013; Diamond and Meehan 2013) and antisocial features (Bateman et al. 2016), the evaluation of treatments for NPD outside of case formulations are nonexistent. Clearly this is a significant gap that psychotherapy research needs to address, and Kernberg has laid the groundwork for us to do just that.

Conclusion

Kernberg has articulated a sophisticated model for how transference-based interventions can be applied to diverse personality presentations. Not only is this model supported by empirical research demonstrating the value of interpretive work for patients with pathological object relations; the specificity with which he describes disparate pathologies provides signposts for future psychotherapy research. This treatment model maintains consistency by grounding itself in a core set of principles, and yet is flexible in its application to a range of transference presentations. For all of these reasons Kernberg’s model occupies a unique place in our field; he has been a pioneer for psychoanalytic theorists and psychotherapy researchers alike.