An Empirical Investigation of Defense Interpretation Depth, Defensive Functioning, and Alliance Strength in Psychodynamic Psychotherapy

Introduction

Defense mechanisms are any of the various, usually unconscious, mental processes that protect the self from shame, anxiety, conflict, loss of self-esteem, or other unacceptable feelings or thoughts. Not all defense mechanisms perform this task in the same manner. Certain defenses (e.g., self-assertion) are more adaptive than others (e.g., projection) by virtue of their ability to resolve conflicts, whereas maladaptive or immature defense may simply exacerbate the issues (Vaillant, 1993). The level of adaptability or defensive functioning of patients is related to psychopathology (Bond, 2004; Bond & Perry, 2004; Bloch, Shear, Markowitz, Leon, & Perry, 1993; Busch, Shear, Cooper, Shapiro, & Leon, 1995; Lingiardi et al., 1999; Perry & Cooper 1989; Spinhooven & Kooiman, 1997; Zanarini, Weingeroff, & Frankenburg, 2009). Moreover, research studies have shown how defensive functioning changes toward a more adaptive level of functioning through psychotherapy (Perry & Bond, 2012; Roy, Perry, Luborsky, & Banon, 2009).

Therapists in psychodynamic psychotherapy emphasize the identification and interpretation of defense mechanisms (Summers & Barber, 2010; Weiner & Bornstein, 2009). It is assumed that interpretive techniques aid patients to develop insight about the defensive process and therefore change them. Interpretations are considered the “fundamental technical instrument” (Etchegoyan, 2005; p. 9) used in psychodynamic psychotherapy and have, as a result, received the most attention in the literature. Despite consensus regarding the importance of interpretations, studies examining the use of this therapeutic technique with defense mechanisms are lacking. This may be due in part to the fact that measures for rating therapeutic technique in psychodynamic psychotherapy only began to appear in the 1980s with instruments like the Psychodynamic Interventions Rating Scale (PIRS: Cooper, Bond, Audet, Boss, & Csank, 2002). A second possible cause for this lack of attention to in the literature defense interpretations was the attention paid to nonspecific factors of psychotherapy process such as therapeutic alliance.

In recent years, several studies specifically examining defense interpretations have emerged (Despland, de Roten, Despars, Stigler, & Perry 2001; Foreman & Marmar, 1985; Hersoug, Hoglend, & Bogwald 2004; Junod et al., 2005; Winston, Winston, Samstag, & Muran, 1993). While the field is still in the initial phases of development, researchers have brought to light the importance of this avenue of enquiry. More specifically, studies of defense interpretation can be divided roughly into two groups: those studies that examined the technique’s relationship to changes in defensive functioning and those studies that examined the technique’s relationship to important process variables such as the therapeutic alliance.

The first to examine these processes were Foreman and Marmar (1985), who found that when a therapist interpreted a patient’s defensive feelings toward the therapist, the alliance tended to improve. Although the study consisted of a sample of only six patients, the authors argued that linking the problematic aspects of the relationship with the therapist to patient defenses in psychotherapy may help to improve outcome. Similarly, Despland et al. (2001) went one step further by developing a system of adjustment where therapist interventions (including interpretations) were matched by ratio to patient defensive functioning so that at every level of patient defensive functioning there was an assumed appropriate level of therapist intervention. Despland et al. (2001) used this ratio to discriminate among low alliance groups, high alliance groups, and therapist-patient dyads with an improving alliance, showing how adjustment predicted membership in one of those three groups. This system of adjustment was, however, met with criticism by Hersoug et al. (2004), who found that what appeared to be an “appropriate” adjustment ratio in some cases was actually associated with either poorer outcome or the wrong assumed alliance group.

Concurrent with this, Junod and colleagues (2005) found that therapist accuracy of defense interpretation was associated with alliance strength. An accurate interpretation was defined as the therapist’s ability to highlight those defensive levels most commonly used by the patient during the session. Junod et al. (2005) determined that higher accuracy of defense interpretation was associated with a stronger therapeutic alliance. Investigating accuracy of defense interpretation highlights the notion that not all interpretations are delivered in the same manner and that characteristics of the interpretations themselves should perhaps be examined as well.

Despite these efforts, much remains to be accomplished, as a number of different aspects of defense interpretations have not yet been investigated. For example, in a review of the literature aiming to identify techniques that are recommended in delivering psychodynamic psychotherapy, Petraglia, Bhatia, and Drapeau (2013) identified a set of ten technical principles, one of which is the need for therapists to consider the “depth” of a defense interpretation, also known as the “surface-to-depth” rule. Fenichel (1941) suggests that analysts proceed in their work from “surface to depth”, that is therapists should not address material that is too far out of the patient’s awareness before he or she is ready to deal with such material. In his review of therapeutic technique in psychodynamic psychotherapy, Etchegoyan (2005) mentions depth of interpretation as an important feature of interpretive techniques. Interpretive depth is understood as the degree to which an interpretation includes elements deeper in the unconscious. Greenson (1967) suggests that the process of describing to the patient the defensive process he or she is using is closer to the surface of consciousness than the history from which the defensive process originated and the motives for which the defense was employed. More recently, Bhatia, Petraglia, de Roten, and Drapeau (2013) conducted a survey of the ten principles put forth by Petraglia et al. (2013) and found that nearly 60% of currently practising psychodynamic therapists rated the “surface to depth” rule as a critical technical principle in dynamic therapy.

Although this principle (consideration of the “depth” of a defense interpretation) is readily identifiable in the theoretical literature, and still widely quoted today in textbooks that describe how to conduct psychodynamic psychotherapy (e.g., Lemma, 2003), a major gap remains in the literature: no study to date has examined depth of defense interpretations empirically. This study attempts to partially address this research gap by examining a number of key variables in dynamic therapy. More specifically, this study aims to explore the relationship between depth of therapist defense interpretations, and patient defensive functioning, and the alliance. The alliance was selected as a key variable because it has been shown to be one of the most robust predictors of outcome (Horvath & Symonds, 1991; Martin, Garske, & Davis, 2000). Specifically, sessions identified as low-alliance and high-alliance were selected as the basis for comparison of defenses and defense interpretations. In addition, because defense interpretations are by definition designed to address patient defenses, session defensive functioning was also examined.

Method

Participants

The sample was collected as part of a larger study of process in psychodynamic psychotherapy at the University of Lausanne, Switzerland (UNIL-EPFL). The present subsample consisted of 36 students between the ages of 18 and 30 years of age (M = 23.94, SD = 3.93). They received one to two sessions per week of manualized (Gilliéron, 1997) Short-Term Dynamic Psychotherapy (STDP), ranging in duration from six months to one year. This treatment is and has already been in use for many years in Lausanne. All participants were outpatients requiring psychiatric or psychotherapeutic assessment at the UNIL-EPFL outpatient clinic. Each received an information document and was given a written informed consent form to fill out. All participants had to be at least 18 years old and present with an anxiety disorder, depressive disorder, or personality disorder that satisfied DSM-IV-TR criteria (APA, 2001). Exclusion criteria included: organic or delirium disorder, substantial alcohol or drug dependence, schizophrenia or other psychotic disorders, bipolar disorders, mental retardation, and antisocial personality disorder. An independent clinician made diagnoses on the basis of a formalized semi-structured interview, using DSM-IV-TR criteria. The participant pool was naturally selected. As such, it closely resembles the manner in which individuals may seek out mental health services in their communities.

The sample is well suited to the present study for several reasons. First, short-term dynamic psychotherapy (STDP) lasting between six months and one year allows sufficient time for defense mechanisms to be expressed by patients in the therapy, and therapists have ample time to identify and interpret patient defense mechanisms. In addition, the length of treatment also allowed for the alliance to go through potential variations of low-alliance and high-alliance processes, which are necessary components of the present research. Furthermore, the present STDP sample also addressed some limitations from previous studies that have investigated interpretation, defense mechanisms, and the therapeutic alliance. Most notably, several studies (Ambresin, de Roten, Drapeau, & Despland, 2007; Despland, de Roten, Despars, Stigler, & Perry, 2001; Drapeau et al., 2008) have examined ultra-brief (four sessions) psychodynamic psychotherapy, which is not representative of the typical manner in which this form of psychotherapy is usually conducted.

Finally, in today’s health care system financial considerations make treatment duration an important consideration and shorter-form therapies have become the norm in many settings (Binder, 2004). Few studies can afford to be conducted on open-ended psychodynamic psychotherapy and the use of STDP can provide researchers with important understandings of the general manner in which psychodynamic psychotherapy unfolds.

Psychotherapists

Psychotherapists for the study consisted of 12 (eight male and four female) clinicians, with more than 10 years of experience with the STDP model. They saw on average three participants each. These psychotherapists also supervised trainees at the center for psychoanalytic psychotherapy (CEPP) at the UNIL-EPFL in Lausanne, Switzerland.

Instruments

The measures for the study were designed to capture three important aspects of psychotherapy: the therapeutic alliance, therapist interventions, and defense mechanisms.

Alliance

The Helping Alliance Questionnaire ([HAq-II] Luborsky et al., 1996) was used to rate alliance strength for individual therapy sessions. The scale shows acceptable levels of convergent validity with other self-rated measures of alliance in use today in psychotherapy research (Luborsky, 2000). For example, the HA-q has been shown to be correlated with the Working Alliance Inventory (r = .74) and California Psychotherapy Alliance Scale (r =.74). Furthermore, the HA-q shows evidence of strong validity (see Luborsky et al., 1996).

For this study, low-alliance and high-alliance sessions were identified based on individual participants’ alliance scores. For example, a low-alliance is defined as a HA-q score one and a half standard deviations below the average HA-q score for that individual participant. Similarly, a HA-q score of one and a half standard deviations below the participant’s mean alliance score were used as the cut-off for the low-alliance session. This method allowed for each participant’s alliance score to be the defining criteria for identifying low-alliance and high-alliance sessions. Sessions identified as either a low-alliance or high-alliance session were included in the present analysis. For the current sample consisting of 36 therapist-patient dyads, there were 45 low-alliance and 39 high-alliance sessions. We attribute the discrepancy to some patients experiencing more than one low-alliance session and high-alliance session. On average, each dyad comprised 1.25 high-alliance sessions and 1.08 low-alliance sessions.

Therapist Interventions

The Psychodynamic Intervention Rating Scale ([PIRS] Cooper et al., 2002) was used to categorize the in-session activities of therapists. It includes 10 types of interventions assigned along a continuum; these interventions are divided into two broad categories: interpretive interventions (defense interpretations, transference interpretations), and supportive interventions (clarifications, reflections, associations, support strategies, questions, contractual arrangements, work-enhancing strategies, acknowledgments). Interpretive interventions (transference and defense) can be further classified into “levels” or depths of interpretation from one to five. Level 1 interpretations point out some mental process (defensive or transference) that the patient is engaging in during therapy, while deeper levels are organized around whether the therapist includes aspects such as motive and historical material in the content of the interpretation. Level 5 is the final and “deepest” interpretation that can be rated using the PIRS. In these cases the rater identifies that the interpretation includes the process used by the patient (i.e. the defense or transference pattern), motive, and the historical origins of this process from the patient’s life. As such, it closely resembles Greenson’s (1967) organization of interpretive techniques in psychodynamic psychotherapy.

The PIRS has shown the ability to discriminate between different tasks in a psychodynamic interview (Perry, Fowler, & Seminuk, 2005) as well as the ability to detect differences in technique use across phases of therapy, including defense interpretations (Hersoug et al., 2003). This is especially important for the present study because only one type of psychodynamic technique (defense interpretations) was examined. The PIRS also takes into account the depth of a specific interpretation that is an important aspect of the proposed study. Hersoug et al. (2005) found that the PIRS interpretive category showed predictive validity and was able pick up changes in maladaptive defense mechanisms in STDP.

Raters in this study were trained to classify the verbal utterances of therapists during psychotherapy sessions by means of verbatim transcripts (according to the interventions described above). When a rater scored an interpretative intervention he or she was also required to note the depth level of the interpretation. Interrater reliability was conducted on 20% of the sample and disagreements were resolved by means of a consensus meeting. The mean intra-class coefficients (ICC 2, 1) for all PIRS categories was .77 (range = .65 to .94; see also Banon et al., 2013).

Defense Mechanisms

Defense mechanisms were assessed using the observer-rated Defense Mechanism Rating Scales ([DMRS] Perry, 1990), in which trained raters assess 30 defenses based on a seven-level hierarchy (see Table 1 below). Each defense level consists of anywhere from three to eight individual defense mechanisms. In addition, an Overall Defensive Functioning (ODF) score, which reflects the average maturity level (1-7) of the patient for a given psychotherapy session, may be computed by taking the weighted mean of each defense by level.

	Level	Defense Mechanism
	Level 7-Adaptive	Self-assertion, Self-observation, Affiliation, Altruism, Sublimation, Suppression, Anticipation, Humour
	Level 6-Obsessional	Undoing, Isolation of Affect, Intellectualization
	Level 5-Neurotic	Reaction Formation, Displacement, Dissociation, Repression
	Level 4-Minor-Image Distortion (Narcissistic)	Idealization of Self/Other, Devaluation of Self/Other, Omnipotence
	Level 3-Disavowal	Denial, Projection, Rationalization, Autistic fantasy
	Level 2-Major-Image Distortion (Borderline)	Splitting of Self/Other, Projective Identification
	Level 1-Action	Acting Out, Passive-Aggression, Help-Rejecting Complaining (HRC)

Enlarge table

The hierarchy used as the basis of the measure is from both theoretical and empirical (Vaillant, 1993; Vaillant, Bond, & Vaillant, 1986) conceptualization of defenses, which places these mechanisms on a continuum from adaptive/mature to maladaptive/immature. The reliability of the DMRS has been well documented in the literature (Perry & Henry, 2004; Perry & Hoglend, 1998; Perry & Kardos, 1995). In general, observer-rated measures are preferred to self-report measures of defensive functioning for psychotherapy process studies (for review see Davidson & MacGregor, 1998; Perry & Ianni, 1998).

Raters were taught to segment and rank participant verbal utterances from transcribed psychotherapy sessions according to the method delineated by the author in the DMRS manual (Perry, 1990). Scoring the DMRS was conducted in two parts. Overall Defensive Functioning is a global measure of overall defense maturity for the participant for a particular session; it is calculated by taking the weighted mean of each defense mechanism scored by level. For example, an ODF = 4.5 would indicate that for a given session, the participant’s average defense falls at the midpoint of level 4 (Narcissistic) and level 5 (Neurotic). The second score for the DMRS involves calculating the proportion of defenses that comprise each level for the participant’s session (please see Table 2 for a breakdown of ODF and level scores for the present sample). This proportion provides a more detailed picture of the defense levels employed most often by the participant. Throughout this process, interrater reliability was calculated for approximately 20% of all transcripts; ICC (ICC 2, 1) for the DMRS of the current sample was published in Kramer, Despland, Michel, Drapeau, and de Roten (2010) and varied between .81 and .95 for defense levels

Results

Preliminary Analyses

A paired t-test was used to compare ODF in the two groups (low-alliance and high-alliance). The multivariate analysis of variance (MANOVA) procedure was used for defense levels one through seven.

No difference in ODF was found between the two groups, t(83) = 1.05, n.s.. The multivariate model for defense levels was also not significant, F(7,76) = .78, n.s.

Interpretation Depth

We examined mean depth of defense interpretation in low-alliance and high-alliance sessions using the non-parametric Wilcoxon Matched Pairs Sign Test. This test was also used to compare the proportion of defense interpretations to total interventions between the two groups (low-alliance & high-alliance). By virtue of being more active in-session, certain therapists may speak more, thereby making more interpretations. Using proportional scores allowed for control of this effect by comparing the ratio of interpretations made as compared to total interventions for the session made by an individual therapist.

Results of the Wilcoxon Matched Pairs Sign Test indicated a significant difference between low-alliance and high-alliance sessions for mean interpretation depth Z(83) = –2.16, p =.03. High-alliance sessions showed less interpretation depth than low-alliance sessions. When proportional scores were used, a non-significant trend was found, Z(83) = –1.92, p =.06, in the same direction. In this analysis, high-alliance sessions showed lower proportions of defense interpretations to total interventions than did low-alliance sessions.

Interpretation Depth and Defensive Functioning

Spearman’s Rank Correlational analysis was used to determine whether there was an association between mean defense interpretation depth (mean depth) and defense functioning (ODF) for each alliance group. A correlation coefficient was also calculated between the total number of defense interpretations and session ODF. Finally, another correlation was run for defense interpretation and DMRS levels one through seven.

Tables 3 and 4 provide a breakdown of the Spearman correlations for interpretation depth and defensive functioning. Mean interpretation depth was significantly correlated to session ODF for low-alliance sessions (r = .30, p = .05) but not for high-alliance sessions (r = .07, n.s.). For defense levels, a significant correlation was found between defense level 3 (disavowal) and mean depth of interpretation for low-alliance sessions (r = .48, p = .00). No other significant correlations were found for defense levels and mean depth for either the low-alliance or high-alliance groups.

Discussion

No evidence was found to the effect that mean defensive functioning is lower in low-alliance sessions as compared to high-alliance sessions. Additionally, there were no significant differences between the two groups of sessions on overall defensive functioning or individual defense levels. These findings suggest that session defensive functioning may be independent of the strength of the therapeutic alliance, which is consistent with the theory that defensive functioning is a stable personality characteristic and little variation is observed when single sessions are examined (e.g., Perry, 2001). These data are also consistent with what has been found empirically, namely a failure to find any relationship between observer-rated measures of defenses and the therapeutic alliance (Hersoug et al., 2002; Siefert et al., 2006).

Results of this analysis were more complex when depth of interpretation was considered. For example, the results suggested that mean depth of defense interpretations in psychodynamic psychotherapy was associated with the quality of the therapeutic alliance for a particular session. In addition, the same therapist may interpret defense mechanisms more “deeply” in a low-alliance session than in a high-alliance session. Since the data are correlational, it is not possible to determine whether a therapist interpreting more deeply causes a weaker alliance or whether the difficulty in the therapeutic alliance compels therapists to make more of these types of interpretations. Despite this, it appears that a relationship exists between the way interpretations are structured in psychodynamic psychotherapy and the strength of the therapeutic alliance.

Interestingly, interpretation depth was significantly related to defensive functioning for low-alliance sessions but not for high-alliance sessions. The association suggests that therapists making deeper defense interpretations had (on average) patients who reported weakened alliances, whereas the opposite was not true for stronger alliance sessions. It may be premature to suggest that dynamically oriented therapists should refrain from making “deeper” interpretations for fear of weakening the alliance but clearly the idea that interpretation depth should be studied in greater detail appears warranted.

Similarly, we found in the analysis some evidence for a relationship between depth of interpretation and defensive levels as measured by the DMRS. In particular, Level 3 or disavowal defenses (projection, denial, rationalization) were moderately correlated (.48) with deeper interpretations made by therapists in low-alliance sessions. The same was not true of high-alliance sessions. Disavowal defenses aim to obscure some aspect of reality to the patient employing them. Although the process of self-deception is common to all defenses, disavowal defenses are predominantly concerned with “refusal” to accept some aspect of reality. As such, the link between this class of defenses and interpretation depth can be viewed as a “mismatch” between the type of information being communicated by the therapist to the patient and the inability of the patient to “hear” or accept this information as an accurate reflection of his or her psychic experience.

There are several points worth exploring related to the use of disavowal defenses as a function of a mismatch between patient and therapist. First, the most obvious function of the disavowal defense is to refuse the content of the interpretation made by the therapist, as noted above. Denial is the most straightforward manner in which the patient can attempt this by flatly “refusing” that which is proposed by the therapist. However, in denial (as opposed to repression) some of the denied material slips into consciousness before it is deemed too conflict laden to be accepted, thereby activating the need for a defense mechanism (Dorpat, 1985). It is important to note that denial is more than simply not agreeing with the interpretation of the therapist, which patients are obviously free to do, but rather signals a “nonacceptance” of unconsciously motivated materials presented to the patient in the form of an interpretation of the patient’s behaviour. Therapists can, of course, be inaccurate and off the mark with their interventions.

The second possible way in which a patient can disavow the content of an interpretation is through the defense of projection. In this case, a patient is unable to use the straightforward denial and must reattribute the content that is disavowed to someone or something else. When this idea is applied to the results of the present study, then the idea emerges that patients in the low-alliance sessions found certain aspects of the deeper interpretations resonating in consciousness but were unable to apply content to themselves. It is possible because it was too conflict-provoking and thus it was ascribed to some other external object.

The third manner in which a patient can disavow in session is through the defense of rationalization. In this case, almost no refusal exists in the patient but rather there is a need to intellectually reduce the effect of the interpretation. For example, the patient reasons and “rationalizes” the interpretation into some diluted product so that it is no longer causing as much intrapsychic conflict or anxiety. In essence, these three defense mechanisms point to the fact that a deep interpretation can push patients to rid themselves of the interpretation, especially if the interpretation itself provokes some intrapsychic conflict. Clinically, patients may be using these defenses to signal that they are either not ready for such content at that particular point in therapy or that they are simply unable or unwilling to accept what the therapist is proposing in his or her interpretive intervention. As a result of the use of these disavowal defenses, the mismatch mentioned above emerges and may result in a low-alliance session. If, however, the therapist deems it necessary to “dig deeper” in such circumstances, by incorporating more and more content into the interpretation (e.g., motive for the defense, link to the past), then theoretically the technical rule of surface to depth is broken, and the therapy itself may be placed in jeopardy. Related to this, Siefert et al. (2006) found that patients with lower ODFs received, in general, more interpretations. While that study did not examine interpretation depth per se, it is consistent with the idea that the use of immature defenses by patients, of which disavowal defenses belong to, may trigger therapists to make deeper interpretations in order to make a bigger impact on what may appear to be, on the surface, a challenge by patients toward the intervention.

It is important to point out that methodologically, those rating defense mechanisms worked independently of those rating therapist interventions; these defenses were not rated as “reactions” to therapist interpretations. Indirect evidence for the idea that a mismatch between therapist intervention and patient defense exists as demonstrated by that the relationship between interpretation depth and disavowal defenses was not found for high-alliance sessions. In essence, therapists who make deep defense interpretations may elicit defense manoeuvres by patients aimed at blocking out the interpretation or denying its relevance.

Vaillant (1994) warned against this therapist-patient dynamic in his book about working with defense mechanisms for patients diagnosed with personality disorders. Vaillant reasoned the potential drawbacks of any kind of interpretive interventions outweigh the benefits when working with this population. He saw the potential for patients feeling misunderstood as being too high. His work represents a departure from classical psychodynamic theory that holds interpretation as the only effective means toward true characterlogical change (Etchegoyan, 2005). Years later Perry and Bond (2005) echoed an idea similar to Vaillant’s (1992) for the same population, but they remained adamant that deep interpretations could be used with these patients. However, the current sample was not diagnosed exclusively with personality psychopathology, thus raising the question of whether deep interpretations carry an inherent risk of making patients feel alienated regardless of the diagnosis. This effect may only emerge with a troubled alliance in non-PD diagnosed patients. It appears that too many deep interpretations in a weak alliance may create a situation in which defenses that are most commonly observed in patients with personality disorders come to the surface.

If therapists do not make the necessary clinical adjustments, patients may feel misunderstood by therapists. Safran and Muran (2000) pointed out how ruptures in the therapeutic alliance are related to premature termination as a product of feeling misunderstood by therapists. While therapeutic ruptures were not per se the focus of this investigation, the present study examined “low-alliance” sessions, which are theoretically similar to a rupture in that the therapeutic alliance is in a state of conflict. It would appear that inappropriate use of defense interpretations may be a possible pathway to help shed light on a disruptive process in the therapeutic alliance.

It is important to note the limitations of the current study. The data associating interpretation depth, defensive functioning, and low-alliance/high-alliance sessions is correlational. As such, it is not possible to high light a causal relationship between these variables. Secondly, the study was conducted on sample of students at the University of Lausanne and may not necessarily be representative of other settings. Thirdly, the sample size of the current study limited power to find all but the most robust findings.

Future studies may potentially expand on the current investigation by incorporating more data points over the course of therapy with every therapist-patient dyad. This way more data could be amalgamated within the complex interactional process that characterizes the psychotherapy relationship. Thus, more research is needed to separate therapist-specific variables as a means of illuminating defenses used by patients in psychodynamic psychotherapy. Ultimately, more needs to be done to understand how therapeutic interventions play a role in patient defense use as well as change in defensive structure.

Conclusion

In conclusion, the empirical investigation of defense interpretation depth appears to be a valuable avenue of study for psychodynamic psychotherapy. Although results were mixed, the ideas gathered from the clinical literature in Petraglia et al. (2013)—that therapists should pay special attention to defense interpretation depth—show some support. Clearly, a great deal more empirical evidence is needed to elucidate how defense mechanisms and interpretations interact within the framework of the therapeutic alliance and to better understand this delicate interplay of forces.