Psychoanalytic Psychology

© 2021 American Psychological Association 2022, Vol. 39, No. 2, 145–153

ISSN: 0736-9735 [Link]

General Patterns in Psychotherapists’ Countertransference

Johan Berg, MSc1 and Lars-Gunnar Lundh, PhD2
1
School of Social Work, Lund University
2
Department of Psychology, Lund University

Countertransference has mainly been studied in psychoanalysis from the perspective of the individual
psychoanalyst, in case reports and theoretical papers, and to some extent in empirical research on a general
level. There is a gap between knowledge about countertransference on the individual and the general level,
as well as between empirical research and clinical practice. The aim of the present study was to examine
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

common patterns in countertransference among individual therapists, using a person-oriented approach. By

This document is copyrighted by the American Psychological Association or one of its allied publishers.

using a cluster-analytic statistical method with data on 219 individual therapists’ countertransference (using
the Therapist Response Questionnaire in relation to a particular psychotherapy), six groups of therapists
with similar patterns in countertransference were identified: (1) over-engaged, (2) disengaged, (3) low
intensity of countertransference, (4) high intensity in countertransference, (5) parental, and (6) sexualizing
countertransference. Some identified patterns could be organized into opposites in dimensions: The level of
intensity (low vs. high), and the experienced distance in the therapy (over-engaged vs. disengaged). The six
patterns may represent prototypical patterns of reaction that could serve as self-knowledge for the therapist
to counter different relational challenges in therapy. Limitations are the use of a cross-sectional design with
only one patient per therapist, and the absence of a formal diagnostic procedure, which precludes any
conclusions about the extent to which these patterns were due to therapist or patient characteristics.

Keywords: countertransference, Therapist Response Questionnaire, TRQ, cluster analysis

When Freud advised psychoanalysts to model themselves “on the During the history of psychoanalysis, the concept of counter-
surgeon, who puts aside all his feelings” (1912, p. 115) and to adopt transference has been defined in several ways, from Freud’s (1910)
an attitude of “emotional coldness” (p. 115), he wrote this both with classical definition, over Heimann’s (1950) totalistic view, to con-
the psychoanalyst’s personal well-being in mind, to ease the burden, temporary conceptions as described by Gelso and Hayes (2007).
and with an intention to create the best possible conditions for Countertransference has mainly been described and studied in case
helpfulness in psychoanalysis. Today, probably few psychoanalysts reports or vignettes in theoretical papers, focusing on the individual
would agree with this recommendation if read literally. It is impos- analyst. Some case reports are bold self-disclosures, and offer a
sible not to be affected by the patient, and even inadvisable not to glimpse into the analyst’s private thoughts, feelings, and sometimes
allow subjective response. The statement is nevertheless important personal wounds or flaws. Case reports have an advantage in the
as Freud recognized that the analyst’s person affects the patient, possibility to gain depth in description and understanding. With only
apart from what is explicitly intended, and this quote is commonly one individual as the study object, there is space for complexity and
considered to refer to what he had called countertransference in
a multitude of aspects. Occasionally, authors have summarized the
another paper (Freud, 1910) even if Freud didn’t use the term here.
clinical experience in reviews of common areas in countertransfer-
In his technical papers, Freud summarized different aspects of what
ence (e.g., Gabbard & Wilkinson, 1994) in the ambition to create
he had learned through practice and believed had general applica-
generalizable knowledge.
bility. Freud (1912) emphasized that any other psychoanalyst, “quite
Unsatisfied with the distinction between the natural sciences and
differently constituted might find himself driven to adopt a different
the humanities, Windelband (1894/1998) made an epistemological
attitude to his patients and to the task before him” (p. 111), and
division between nomothetic and idiographic sciences. The nomo-
thereby recognized the possibility of individual differences in both
thetic sciences (e.g., physics) seek stable principles, whereas the
technique and countertransference. Consequently, Freud acknowl-
idiographic sciences (e.g., history) focus on the particular and
edged a span from general directions of psychoanalytic technique to
unique, seeking understanding or explanation of single instances.
a recognition of adaptations on the individual level.
It is debatable whether psychology should mainly be viewed as a
nomothetic or idiographic science, or a combination (Lundh, 2015;
Wallerstein, 2009), but psychoanalysis relies heavily on the idio-
graphic and there is a relative lack of nomothetic or empirical studies
This article was published Online First September 27, 2021. (Bornstein, 2007; Lingiardi et al., 2016). This is problematic for
Johan Berg, MSc [Link]
psychoanalysis due to its ambitions of generalizability, not only to
Lars-Gunnar Lundh, PhD [Link]
We have no known conflict of interests to disclose.
understand idiosyncratic states or processes. The gap between
The first author was supported by Bertil Wennborgs stiftelse. clinical practice and empirical research, as well as between idio-
Correspondence concerning this article should be addressed to Johan graphic and nomothetic perspectives needs attention. At the same
Berg, MSc, School of Social Work, Lund University, P.O. Box 23, Lund time, there seems to be reason for a moderate optimism concerning
22100, Sweden. Email: [Link]@[Link] the clinical usefulness of empirical studies in psychoanalysis

145
 146 BERG AND LUNDH

(Lingiardi et al., 2016), and conversely fruitful for psychotherapy Inclusion/Exclusion Criteria
research to meet the idiographic perspective. One way of narrowing
To be able to perform all statistical analyses (i.e., both cluster
these gaps is by adopting a “person-oriented approach” with a
analysis and logistic regression analysis) on the same sample, a
purpose of studying patterns at the level of the individual
selection was made with the largest groups of therapists (i.e., only
(Lundh, 2019). Traditional empirical research in psychology as
cognitive-behavioral or psychodynamic therapists), and the three
well as in psychotherapy has focused on how scores on variables
largest diagnostic groups, that is, mood disorders (F30–F39), anxi-
correlate or differ between groups in a sample (variable-oriented
ety, stress-related and somatoform disorders (F40–F48), and per-
research), but there are also other statistical methods (e.g., cluster
sonality disorders (F60–F69). This excluded therapists with
analysis) that allow studies of individual patterns and the classifica-
theoretical orientation “eclectic/integrative” (n = 20), “family-
tion of these (in person-oriented research). Cluster analysis is used
oriented” (n = 7), and “other” (n = 1), and also patients with
to create classifications of individuals with similar patterns, by
psychoactive substance use (n = 4), psychosis (n = 1), eating
creating as homogeneous groups as possible with the individuals
disorder (n = 3), autism or Asperger syndrome (n = 4), and patients
that resemble each other in a specific set of parameters. For example,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

without stated psychiatric diagnosis (e.g., personal therapy;

factor analysis can reveal patterns among variables or items in a
This document is copyrighted by the American Psychological Association or one of its allied publishers.

n = 12). The excluded categories were considered too small to

questionnaire, whereas a cluster analysis examines patterns with the
be meaningfully included in the subsequent regression analysis.
person as the organizing principle. A cluster analysis places different
Hence, a selected, more homogenous sample was created, resulting
individuals with similar patterns into relatively homogeneous
in N = 219.
groups, whereas a variable-oriented factor analysis, for example,
looks for similarities (e.g., covariance) between the items in a
questionnaire. Participants
As to countertransference, there are factor-analytical studies using
questionnaires, such as the Therapist Response Questionnaire Basic demographics and professional characteristics about the
(TRQ; Betan et al., 2005), that suggest different categories of therapists are shown in Table 1. The therapists had practiced
countertransference (Berg et al., 2019; Betan et al., 2005; psychotherapy for an average of 15.4 years (SD = 10.3, range =
Tanzilli et al., 2016). Berg et al. (2019), for example, identified 2–55), and reported a current workload at an average of 22.1
the following seven factors: Helpless/Inadequate, Overwhelmed/ weekly therapy sessions (SD = 9.2, range = 1–48). Due to the wide
Disorganized, Hostile/Angry, Parental/Protective, Disengaged,
Special/Overinvolved, and Sexualized. To our knowledge, no
cluster analysis using these factors has been performed. A cluster Table 1
analysis can identify groups of therapists with similar patterns in Therapist Demographics and Professional Characteristics
countertransference.
The purpose of the present study was to identify patterns of Present study
National statisticsa
countertransference, defined as the therapist’s feelings and reactions Therapist characteristic n % %
toward the patient (i.e., the totalistic view), by applying cluster
analysis with a sample of psychotherapists using TRQ. We also Gender
Female 164 74.9 73.9
explored if the patient’s diagnosis, the therapist’s gender, therapist’s Male 55 25.1 26.1
workload, theoretical orientation or length of therapy could predict Age
their countertransference. We expected to find common and general 30–39 22 10.0 2.0
patterns that unite therapists in clusters. 40–49 69 31.5 14.2
50–59 72 32.9 30.4
60–69 47 21.5 41.9
70– 9 4.1 11.4
Method Profession
Psychologist 113 51.6 43.5
Sampling and Procedure Social worker 59 26.9 c

Nurse 16 7.3 6.9

Therapists in Sweden were asked to participate in a validation Psychiatrist 5 2.3 6.6
study of therapists’ reactions in therapy. Various channels were used Otherb 26 11.9
in the sampling, including asking four national organizations for Theoretical orientation
psychotherapists or psychoanalysts to use their email lists (two for PDT/PA 124 56.6
CBT 95 43.4
cognitive-behavioral therapists, one for psychodynamic therapists Work setting
and one for psychoanalysts, in total approx. 3,000 emails), Face- Private practice 112 51.1 53.9
book groups for psychologists, and supervisors at universities Public sector 104 47.5 46.1
training psychotherapists. The sample is the same as used in a Other work setting 3 1.4
previous study (Berg et al., 2019). A webform was used to collect Note. PDT/PA = psychodynamic therapy/psychoanalysis. CBT = cognitive-
information from the individual therapist. The therapists were behavioral therapy.
instructed to fill in TRQ having in mind the last patient in psycho-
a
Statistics on professionally active licensed psychotherapists (N = 4,981) in
therapy the previous working week. The therapists were not paid to Sweden 2016, retrieved April 11, 2020 from the Health and Welfare Statistical
Database: [Link] b Occupational
participate. No identifying personal data was collected. According therapists, physiotherapists, or other therapists with an accepted degree on
to Swedish regulations, this kind of study is exempt from institu- the doctorate, master or bachelor level. c Social workers are not presented
tional review. separately in the national database of licensed health care practitioners.
 GENERAL PATTERNS IN COUNTERTRANSFERENCE 147

sampling procedure with presumably overlapping email lists and Demographics and Other Information
Facebook groups, it was not possible to estimate the response rate.
However, the distribution of gender and work settings conformed The therapists gave some basic demographic information about
with available national data (see Table 1), but with some overrep- themselves (age, gender), working site (private practice or employed
resentation of psychologists and younger therapists, probably due to as a therapist, i.e., public sector), profession (e.g., psychologist,
used sampling procedure. psychiatrist, social worker, nurse), principal theoretical orientation
The psychotherapies were ongoing and had lasted for an average (e.g., cognitive-behavioral, psychodynamic, eclectic), number of
of 35.0 sessions (SD = 76.8, range = 1–800) when the therapist years of psychotherapeutic experience, and number of hours with
filled out the form. Only therapists contributed with data (i.e., no weekly psychotherapeutic practice with patients. They further re-
patients were involved). Table 2 shows patient demographics and ported the age and gender of the patient, the current number of
diagnostic information. sessions in the therapy, and the patient’s diagnosis. The therapists
were not required to use any specific diagnostic procedure and the
diagnosis could be expressed in plain language or diagnostic code.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Measures The stated diagnosis was converted to the closest appropriate Inter-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

national Statistical Classification of Diseases and Related Health

Therapist Response Questionnaire Problems 10th Revision (ICD-10) category by an external experi-
enced research psychiatrist and the first author separately, and all
TRQ (Betan et al., 2005) is a 79-item questionnaire designed to
ratings were compared and discussed until consensus was reached.
measure various feelings, reactions, and responses a psychother-
apist can have in psychotherapy with a particular patient. The
therapist is instructed to try to describe the way they have felt with Statistical Analysis
the particular patient over the course of the entire therapy, so far.
Cluster Analysis
The questionnaire is used by the psychotherapist alone, with no
involvement of the patient. The items are selected to capture a Cluster analysis was used as a method for grouping the therapists
wide range of manifestations of countertransference, from rela- into different categories according to their profiles of the standardized
tively simple feelings (e.g., “I feel angry at him/her”) to more scores on the seven TRQ factors. We followed a procedure in four
complex situations (e.g., “I have to stop myself from saying or steps to find an appropriate cluster solution with the statistical software
doing something aggressive or critical”), from relatively common ROPstat, Version 2.0 (Varga et al., 2015). First, because it is generally
feelings (e.g., “I feel like I understand him/her”) to presumably known that all cases are not easy to fit within well-defined categories,
rarer situations (e.g., “I tell him/her I love him/her”). The items the residue module in ROPstat was used to identify and exclude 16
are written in everyday language, without explicit theoretical outliers (defined as not having any “twin” within an Average Squared
assumptions, easily recognizable for the clinician. Each item is Euclidian Distance, ASED of 0.7). One case was identified as having
rated on a 5-point Likert scale, ranging from 1 (not true) to 5 (very an identical factor score as another and was therefore excluded from
true). The Swedish translation of TRQ was used, standardized by the cluster analysis; n = 202 were left for clustering. Second, Ward’s
Berg et al. (2019), and scores were calculated on seven TRQ hierarchical clustering method was applied, which starts by consider-
factors: Helpless/Inadequate, Overwhelmed/Disorganized, Hos- ing each individual case as a separate cluster; at each subsequent step,
tile/Angry, Parental/Protective, Disengaged, Special/Overin- the two clusters are then merged that result in the smallest decrease in
volved, and Sexualized. The scales constructed on the basis of the explained error sum of squares (EESS). The purpose of this is to
these factors had an acceptable internal consistency (Cronbach’s provide as homogenous clusters as possible, and the process continues
α .68 ≤ α ≤ .87; Berg et al., 2019). until the optimal cluster solution is identified. Thirdly, the optimal
number of clusters was decided using three criteria described by Varga
et al. (2015): (a) The cluster solution should have an EESS% well
Table 2 above 50% and show a substantial drop to a lower level at the next
Patient Demographics and Diagnosis fusion step. (b) The individual clusters should be as homogeneous as
Patient characteristic and diagnosis N %
possible, preferably with a homogeneity coefficient (HC) < 1 per
cluster. However, as all but one factor were rather skewed (Berg et al.,
Gender 2019), a higher HC was expected for most clusters. (c) The clusters
Female 162 74.0 should be theoretically meaningful. The six-cluster solution was
Male 56 25.6
Other 1 0.5 chosen as preferred according to these criteria. Fourthly, a nonhierar-
Age group chical relocation procedure (Wishart, 1987) was carried out in order to
17–24 34 15.5 improve the homogeneity of the clusters and to increase the variance
25–39 112 51.1 explained by the cluster solution. Homogeneity improved by this with
40–59 65 29.7
an increase of EESS% from 52.46 to 54.62.
60– 8 3.7
Diagnosis (ICD-10)
Mood disorder (F30–F39) 83 37.9 Logistic Regression
Anxiety, stress-related, and somatoform 106 48.4
disorder (F40–F48) A series of binary logistic regressions (SPSS Version 27) were
Personality disorder (F60–F69) 30 13.7 used to assess the impact of patient’s diagnosis, the length of the
Note. ICD-10 = International Statistical Classification of Diseases and therapy so far, therapist’s workload, gender, and theoretical orien-
Related Health Problems 10th Revision. tation on the likelihood of belonging to each of the various clusters
 148 BERG AND LUNDH

(coded as dummy variables). To reduce the risk of using too many The major characteristic is that the therapists are overinvolved,
predictors in the logistic regression (“overfitting”), the analysis was making deviations from the usual frame and finding themselves
performed in two steps for every cluster: First, a bivariate analysis transgressing their usual limits in the contact with the patient (TRQ
with all potential predictors, and subsequently the logistic regression factor Special/Overinvolvement). There is also an experience of
with only those predictors that passed the threshold of p < .10 in the being helpless or insufficient in the contact with the patient (elevated
bivariate analysis. The Hosmer–Lemeshow goodness-of-fit was not scores on the TRQ factor Helpless/Inadequate), as well as experienc-
significant for any cluster, and by that indicated a good fit to ing annoyance and angry feelings toward the patient (elevated
the model. scores on the TRQ factor Hostile/Angry). This is finally, to a lesser
degree, mixed with warm and nurturant feelings, in a parental way
Results (TRQ factor Parental/Protective). Bivariate analysis and subsequent
logistic regression with patient’s diagnosis, therapist’s gender,
Figure 1 displays in graphical form the resulting six clusters using theoretical orientation and workload, and therapy length as pre-
z-scores for each factor index (showing the number of standard dictors did not show any significant impact on membership in this
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

deviations over or under the group mean for each factor). Three cluster.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

clusters are characterized by a single elevated factor, two clusters

show a generally high respectively low level in the factor scores, and
one cluster is a mixture of elevated factors. The description of the Cluster 2: Disengaged Countertransference
therapists under the following cluster headings are based on the
items in the questionnaire. Some cluster labels have potentially This group, n = 41 (18.7%), HC = 1.13, of therapists feel dis-
negative connotations, but at this stage we do not know if the engaged or bored with the patient (elevated scores on the TRQ factor
specific countertransference is harmful or not. Disengaged). There is a feeling of distance to the patient with the
therapists reducing their engagement in the particular therapy. This
is an active experience of being disengaged, not a generally low
Cluster 1: Over-Engaged Countertransference
score on countertransference as seen in Cluster 3. None of the
The first cluster, n = 22 (10.0%), HC = 1.43, contains therapists potential predictors showed any significant impact on membership
who describe different aspects of over-engagement in the patient. in this cluster.

Figure 1
Clusters and Patterns of Average z-Scores Across the Seven Factor Indexes

Note. F1 = helpless/inadequate; F2 = overwhelmed/disorganized; F3 = hostile/angry; F4 =

parental/protective; F5 = disengaged; F6 = special/overinvolved; F7 = sexualized.
 GENERAL PATTERNS IN COUNTERTRANSFERENCE 149

Cluster 3: Low Intensity of Countertransference Discussion

The third cluster, n = 59 (26.9%), HC = 0.43, contains therapists The present study shows that it is possible to find coherent,
who are reporting a generally low intensity of countertransference. interpretable, meaningful, and sufficiently distinct clusters of psy-
These therapists not only report a low degree of sexual and chotherapists categorized in terms of their countertransference
aggressive feelings, and a low degree of warm or parental attitudes, patterns. There were six groups of psychotherapists that showed
but also a low degree of feeling overinvolved as well as low degree distinguishable patterns of countertransference and one residual
of boredom and inattention in relation to their patient. The latter group of therapists who could not be categorized. The different
clearly differentiate them from the therapists in Cluster 2. The groups were therapists with (1) over-engaged countertransference,
cluster is relatively large and homogenous. These therapists were (2) disengaged countertransference, (3) low intensity of counter-
more likely to have a cognitive-behavioral than a psychodynamic transference, (4) high intensity of countertransference, (5) parental
orientation (OR = 3.82, p = .001), and tended to work more hours countertransference, and (6) sexualized countertransference.
with psychotherapy per week (OR = 1.042, p = .022). It is worth noting that some clusters represent opposite aspects
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and could possibly be regarded as thematic dimensions in counter-

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Cluster 4: High Intensity of Countertransference transference. Two of the clusters seem to be about the difficulty of
finding a therapeutically optimal distance to the patient (Gabbard &
The fourth cluster, n = 13 (5.9%), HC = 2.06, consists of thera- Wilkinson, 1994), either by being too enmeshed with the patient
pists that experience a high intensity and wide range of counter- (Cluster 1, over-engaged countertransference) or by becoming less
transference reactions and feelings in many areas (TRQ factors engaged by the patient (Cluster 2, disengaged countertransference).
Helpless/Inadequate, Overwhelmed/Disorganized, Hostile/Angry, A possible “trait”-like understanding of the therapist could involve,
Disengaged, and Special/Overinvolved) regarding the contact for example, Balint’s (1955) contrast between the philobatic person
with the patient. The feelings or responses are strong almost (who prefers distance, independence, self-reliance, and avoids
regardless of the nature of countertransference, undifferentiated, hazardously close contact with objects) and the ocnophilic person
difficult to conceptualize. The therapists’ reactions and feelings (who prefers closeness and avoids horrific empty spaces), or the
seem to lack systematic pattern and can potentially display a rather contrast between a dismissing and preoccupied attachment style
chaotic nature, but are at the same time neither parental, nor (Main, 2000). But these patterns might also represent more tempo-
sexualizing. None of the potential predictors showed any significant rary adaptations to patients with specific problematic forms of
impact on membership in this cluster. relating, or more complex forms of interactions between therapist
and patient, including a degree of therapist identification with the
patient. Gelso and Hayes (2007), for example, describe how both
Cluster 5: Parental Countertransference
over- and under-identification with the patient can result from, and
This cluster, n = 48 (21.9%), HC = 0.72, contains the therapists lead to, countertransference issues. If the therapist overidentifies
who experience predominantly warm, parental, nurturant, compas- with the patient, the optimal distance could be lost, and the therapist
sionate, or protective feelings toward the patient (TRQ factor is at risk of becoming overinvolved in the patient, were “deep
Parental/Protective). There is an absence of unpleasant feelings empathic attunement becomes confused with enmeshment” (Gelso
or negative action impulses toward the patient. The patient makes & Hayes, 2007, p. 68). Similarly, if the therapist underidentifies
the therapist feel good about themselves. The cluster is rather large with the patient, this could lead to or result from a distance to or
and homogenous. None of the potential predictors showed any rejection of the patient.
significant impact on membership in this cluster. It would probably be a simplification to treat these countertrans-
ference patterns as either exclusively due to patient communication
(e.g., projective identification; e.g., Kernberg, 1965) or therapist
Cluster 6: Sexualized Countertransference
traits. A more reasonable hypothesis is to assume an interaction
The last cluster, n = 20 (9.1%), HC = 1.29, contains therapists between the push-and-pulls in the psychotherapy, and the therapist’s
with an elevated level of sexualizing responses toward their patient personal inclination to retreat to either distance or closeness when
(TRQ factor Sexualized). This is a group of therapists who experi- put under pressure, according to the therapist attachment style
ence generally more than average sexual attraction or tension with (ambivalent/hyperactivating or avoidant/deactivating attachment)
the patient or notice themselves being flirtatious toward the patient. or other therapist personality traits (e.g., impatience, difficulties
Bivariate analysis and subsequent logistic regression showed that with aggression, or wanting to “give more” to the patient in need
the therapists in this cluster were more likely male than female to gain approval).
(OR = 3.88, p = .006). Two other clusters also seem to be related as opposites. Cluster 3,
low intensity of countertransference, is characterized by a generally
low level of feelings and reactions, whereas Cluster 4, high intensity
Residual Group
of countertransference, indicates a high level of countertransference
The residual group, n = 16 (7.3%), contains therapists that could in most areas. The therapists in the low-intensity group did not report
not be categorized in any of the above clusters. The therapists’ much of either pleasant or unpleasant emotional responses or
patterns of countertransference did not match any of the above behaviors. In the high-intensity group, the pattern was the opposite
groups but were more individually unique (idiosyncratic). None of with elevated responses in most countertransference areas. Obvi-
the potential predictors showed any significant impact on member- ously, a high intensity of countertransference feelings or reactions
ship in the residual group. could imply a potentially harmful situation, but a low intensity of
 150 BERG AND LUNDH

countertransference is also at risk as this could signal a denial or Money-Kyrle (1956) considered the parental countertransference as
suppression of problematic or painful experiences in the therapist the “normal countertransference” and as a basic condition for
(Gelso & Hayes, 2007). psychoanalysis. If countertransference is defined as “the total
Turning to the individual clusters, one consists of therapists who emotional reaction of the psychoanalyst to the patient”
in various ways experience themselves as disengaged, indifferent, (Kernberg, 1965, p. 38), this should include parental feelings as
bored or uncommitted in the contact with the patient (Cluster 2, countertransference. However, in the original Freudian conception
disengaged countertransference). This reaction is contrary to the of countertransference (Freud, 1910), or the more recent definition
low-intensity group by involving an active experience of lack of made by Gelso and Hayes (2007), who attribute countertransference
commitment, not a general lack of feelings or reactions. Disengage- to the therapist vulnerabilities and blind spots, these positive
ment is sometimes described as a component in burnout (Maslach & parental feelings should probably not be considered countertrans-
Leiter, 2016), and if the countertransference pattern of disengage- ference. Still, a positive, parental feeling for the patient could as well
ment is connected to the individual therapist and not the particular imply that the therapist has an inclination for almost always entering
therapy, the pattern could be an indication of a potential burn-out primarily supportive or nurturing roles in therapeutic relationships,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

situation. Westwood et al. (2017) found a high prevalence of regardless of the patient’s needs. In some cases, this could be due to
This document is copyrighted by the American Psychological Association or one of its allied publishers.

burnout with disengaged feelings in the therapists in the British the therapist’s narcissistic wishes to be loved or idealized by
program Improved Access to Psychological Therapies (IAPT), patients. This could create psychotherapy stalemates as the parental
especially connected to organizational–environmental variables as therapist could avoid challenging themes (e.g., aggression, destruc-
workload, administrative duties, lack of sufficient supervision, and tiveness, envy, sexualization, seduction, shame). McClure and
so forth. It is, therefore, possible to speak of context-induced Hodge (1987) found that if the therapist strongly likes the patient,
countertransference, and perhaps broaden the concept outside the the therapist has tendencies to overestimate similarities, that is to
strict patient–therapist interaction. regard the patient’s personality much too similar to the therapist’s;
Disengagement is at risk of being malign countertransference. In and if the therapist dislikes the patient there will be a corresponding
a case study, Ulberg et al. (2014) showed a difference in treatment tendency to overvalue the differences in personality. Strong parental
outcome when comparing a high to a low disengaged therapist, and feelings could thus distort the picture of the patient.
Dahl et al. (2017) further studied the adverse effect of therapist Of relevance here are also Dahl et al.’s (2014) study of interaction
disengaged feelings in psychotherapeutic work with transference, effects between the therapist parental feelings (measured by the self-
especially for patients with poorer quality of object relations. It report instrument Feeling Word Checklist), the patient’s personality
seems advisable to avoid interventions based on the transference (in disorder and the use of transference interpretations. When parental
contrast to focusing on interpersonal relations outside therapy) when feelings were lower, transference interpretations were beneficial for
disengaged feelings are elevated, at least until the disengaged all patients regardless of the level of personality disorder. If parental
feelings are managed. feelings, on the other hand, were stronger, patients with more
The disengagement could be a therapist blind spot as every other symptoms of personality disorder had more benefit from interpreta-
pattern, but Tanzilli et al. (2016) found small but significant correla- tions focusing the relation to the therapist (here-and-now transfer-
tions between the disengaged index in TRQ and a patient personality ence interpretations), whereas for patients with fewer signs of
disorder diagnosis of a schizoid, schizotypal, narcissistic, or obsessive personality disorder, focus on interpersonal relationships outside
personality. Therefore, it is reasonable to investigate what parts in the of psychotherapy (extra-transference interpretations) were more
therapeutic interaction that might contribute to this pattern. favorable.
There were also a group of therapists with the opposite charac- Finally, there was a cluster of therapists with sexualized responses
teristic that is therapists who report over-engagement in their or feelings toward the patient (Cluster 6, sexualized countertrans-
patients (Cluster 1, over-engaged countertransference). A certain ference). Men are more likely than women to appear in this group. It
amount of ambition and engagement is necessary for the practice of was probably various kinds of sexual trespasses among the early
psychotherapy or psychoanalysis, but there are risks when becom- psychoanalysts that prompted Freud to warn against countertrans-
ing too involved or enmeshed, and thereby lose perspective. This ference (Britton, 2003). Sexualized, erotic or romantic countertrans-
could be an aspect of “furor sanandi” that Freud (1915, p. 171) ference is undeniably complicated for the psychotherapist but seems
warned against, a too strong “passion for curing people” or rage to all the same to be a common phenomenon (Pope et al., 2006) and
cure, as it is sometimes understood. There are dangers connected to probably underreported in surveys due to its delicate nature. Not
therapeutic ambition or over-eagerness, or when the therapist everyone is as frankly self-revealing as Searles (1959) when he
becomes impatient for therapeutic success. stated that he had found “time after time, that in the course of the
A further cluster is the therapists who experience themselves as work with every one of my patients who has progressed to, or very
parental or feeling protective toward the patient (Cluster 5, parental far toward, a thoroughgoing analytic cure, I have experienced
countertransference). These therapists reported pleasant feelings of romantic and erotic desires to marry, and fantasies of being married
warmth, affection, or compassion with the patient, but also sadness to, the patient” (p. 180).
or anger toward other people affecting the patient’s life. The There is a wide spectrum between the behaviors that Freud
therapeutic situation often invites such feelings. It is easy prima reacted to and the feelings that Searles describes. It is customary
facie to associate this with something natural or positive per se, or in psychoanalytic literature to differentiate between two levels of
with a prerequisite for psychotherapy as Rogers (1957) suggests sexualized transference, that of erotic and of erotized transference
with the concept of “unconditional positive regard” (even if (Blum, 1973; Bolognini, 1994), with the latter denoting a more
this concept is not equivalent to parental countertransference). It regressive variant, ego-syntonic, and with an absence of the “as-if”
is not even clear that this should be labeled countertransference. aspect of transference. In congruence with this, Gabbard and
 GENERAL PATTERNS IN COUNTERTRANSFERENCE 151

Wilkinson (1994), as well as Bonasia (2001), propose a counter- one of these countertransferential prototypes is essential to counter
transference subgroup, the erotized, when the “as-if” quality for the different relational challenges as well as utilizing the beneficial
therapist is lost, and the feelings are experienced as real, concrete, potential of countertransference work. Knowledge about these
and possible to realize. A questionnaire, such as TRQ, cannot make general countertransference themes is also vital for psychotherapy
this differentiation, so this cluster could contain both therapists with supervisors and is applicable regardless of whether the psychother-
erotic as well as erotized countertransference. The items in this area apy has an explicit relational approach (e.g., two-person psychol-
seem to be more drawn to the therapist’s experience/feelings, than ogy) or is oriented to a more directive or interpretative approach.
actual behavior. The erotized variant has probably a higher degree of
malign potential (Gabbard, 2017), and is more likely to create a
Limitations and Future Directions
blind spot for the therapist (e.g., due to narcissistic vulnerability).
The management and differentiation of erotic and erotized counter- There are several limitations to the present study. First, the
transference is dependent on honest self-scrutiny and candid con- absence of a more structured diagnostic procedure, and other
sultation with a trusted colleague. psychological data about the patients, limits the possibility to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

There was also a residual group. This is often considered as evaluate the patients’ contribution to the countertransference. Future
This document is copyrighted by the American Psychological Association or one of its allied publishers.

outliers, but in person-oriented research it should probably more studies with TRQ could address this by including more reliable
correctly be viewed as a group consisting of therapists with rarer diagnostic measures and other instruments to measure patient
patterns of countertransference, for example, a more unusual com- characteristics and functioning.
bination of several elevated factors. Second, the cross-sectional design only provides a momentary
Membership in two clusters could be predicted from background picture of the therapists’ responses to their patients, and we do not
variables. Besides the above-mentioned influence of therapist’s know anything about the degree of stability and change in their
gender, the therapists with low intensity of countertransference countertransference patterns over time. A longitudinal design with
were more likely working with cognitive-behavioral therapy repeated measures may be used in future research to address this
(CBT) than with psychodynamic. This difference could be under- question.
stood in several ways. For example, it could be attributed to A third limitation is that the sampling procedure was such that no
differences in training (more or less focus on countertransference accurate response rate could be calculated. It would have been
issues), method (different methods induce therapist feelings or preferable to have a random sample of psychotherapists with a good
reactions in varying degrees), patient populations, or therapist’s response rate. Fourth, as we used a self-report questionnaire, we
personality (different therapeutic orientations attract different kinds could not know the extent of influence from potentially unconscious
of therapists). The nature of the present data, however, does not aspects of the countertransference. Fifth, as some individual clusters
allow any attribution of these effects to specific factors. The therapist were small, there is a potential issue of “overfitting” the model in the
workload also turned out to be a significant predictor for therapists logistic regression, and therefore the resulting differences between
with low intensity of countertransference. One possible explanation the clusters should be interpreted with caution.
could be that therapists with a high workload are at risk of becoming Finally, it might be questioned whether the identified patterns of
emotionally numb and that this may increase the likelihood of countertransference can rightly be labeled “general,” as the original
reporting fewer feelings or reactions. sample was reduced by excluding therapists with other orientations
The patient’s diagnosis of personality disorder has elsewhere than psychodynamic and CBT, as well as therapies with patients
(Betan et al., 2005; Tanzilli et al., 2016) been shown to correlate outside the three largest diagnostic groups. This reduction of the
significantly with various factors in TRQ, but in the present study sample was made to allow analysis of all statistics on the same
the patient’s diagnostic category did not predict cluster membership homogenous selection. This resulted in a reduction of the sample by
in any cluster. This could be due to sample differences, as the present 18%, and the results could potentially differ from the complete
study focuses on a wider range of patient psychopathology without sample. However, a cluster analysis using the original sample
any differentiation between the personality disorders, or to too large (N = 268) showed in essence the same clusters (with an addition
heterogeneity in the used diagnostic groups. of a cluster with distancing countertransference, and with the current
To conclude, nomothetic research has a potential to inform the over-engaged cluster as a singular overinvolved factor). This in-
individual clinician, and conversely, the idiographic perspective can dicates a rather stable cluster structure and makes it reasonable to
also be a fruitful unit in empirical research. We believe that a person- hypothesize that the identified patterns may be general in nature.
oriented approach to research is a promising perspective. For Similar analyses with other representative samples of therapists,
example, the present cluster analysis suggested three clusters of however, are needed to corroborate this hypothesis.
therapists, characterized by elevated scores on singular factors from There are also other interesting questions that could be made
the previous factor analysis (Berg et al., 2019), and in addition to subject to future research. For example, the present results say
this, three new patterns characterized by specific combinations nothing about the degree, or under which conditions, the identified
among factors (representing 43% of the therapists). A variable- patterns of countertransference are harmful or beneficial. This could
oriented approach using for example only factor analysis could not be investigated by including TRQ in a psychotherapy out-
detect the latter groups of therapists. come study.
We believe that general patterns in countertransference are central Another development could be to let therapists use TRQ repeat-
elements in the therapist’s or analyst’s self-knowledge, which could edly with different patients and to compare the statistical variance in
be gained in the therapist’s personal therapy, the analyst’s training their answers. A relatively high variance in a particular scale could
analysis, and in supervision. Understanding how the individual indicate that the patterns represent more of temporary states due to
clinician may experience and act in accordance with each and every the specific patient involved, whereas a relatively lower variance
 152 BERG AND LUNDH

suggests that the answers are stable over patients and represent more Dahl, H.-S. J., Røssberg, J. I., Crits-Christoph, P., Gabbard, G. O., Hersoug,
of a therapist’s trait. A. G., Perry, J. C., Ulberg, R., & Høglend, P. A. (2014). Long-term effects
Yet another question is to what extent the use of questionnaires of analysis of the patient-therapist relationship in the context of patients’
such as the TRQ can be used for educational purposes as a part of the personality pathology and therapists’ parental feelings. Journal of Con-
development of therapist self-knowledge and helpful in supervision sulting and Clinical Psychology, 82(3), 460–471. [Link]
a0036410
of psychotherapists, and whether they can be used to provide
Freud, S. (1910). Future prospects of psychoanalytic therapy. The standard
therapists with feedback that can facilitate the therapeutic process
edition of the complete works of Sigmund Freud (Vol. XXI, pp. 139–151).
and improve the treatment outcome. Lastly, we do not know much Hogarth Press.
about the phenomenology of the respective countertransference Freud, S. (1912). Recommendations to physicians practising psycho-analy-
pattern, such as how the therapists experience, think about, interpret, sis. The standard edition of the complete psychological works of Sigmund
or manage the different kinds of countertransference pattern. This Freud (Vol. XII, pp. 109–120). Hogarth Press.
could be made subject in future qualitative research. Freud, S. (1915). Observations on transference-love: Further recommenda-
tions on the technique of psycho-analysis III. The standard edition of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

complete psychological works of Sigmund Freud (Vol. XII, pp. 157–171).

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Hogarth Press.
摘要 Gabbard, G. O. (2017). Sexual boundary violations in psychoanalysis: A 30-
year retrospective. Psychoanalytic Psychology, 34(2), 151–156. https://
精神分析中的反移情主要从个体精神分析师的角度、案例报告和理 [Link]/10.1037/pap0000079
论论文的角度,在一定程度上的一般层面的实证研究中进行。在关于 Gabbard, G. O., & Wilkinson, S. M. (1994). Management of countertrans-
个人和一般层面的反移情的知识之间存在着一个缺口,正如在实证研 ference with borderline patients. American Psychiatric Press.
究和临床实践之间的缺口。本研究的目的是使用以人为导向的方法, Gelso, C. J., & Hayes, J. A. (2007). Countertransference and the therapist’s
研究个体治疗师中的反移情的常见模式。通过使用集群分析统计方 inner experience: Perils and possibilities. Lawrence Erlbaum. https://
法,提供219名个体治疗师的反移情数据(使用治疗师对特定心理治疗 [Link]/10.4324/9780203936979
的回应问卷),确定了6组在反移情中具有类似模式的治疗师: (1)过度 Heimann, P. (1950). On counter-transference. The International Journal of
参与的、(2)脱离的、(3)反移情强度低、(4)反移情高强度、(5)父母 Psycho-Analysis, 31, 81–84.
式的以及 (6)性欲化反移情。一些已识别的模式可以组织为对立的维 Kernberg, O. (1965). Notes on countertransference. Journal of the American
度:强度水平(低对高),以及治疗中体验的距离(过度参与对脱离)。这 Psychoanalytic Association, 13(1), 38–56. [Link]
六种模式可能代表典型的反应模式,可以作为治疗师的自我认识,以应 000306516501300102
对治疗中不同的关系挑战。限度是横截面设计的使用,每个治疗师只 Lingiardi, V., Holmqvist, R., & Safran, J. D. (2016). Relational turn and
有一个病人,以及缺乏正式的诊断程序,这排除了任何关于这些模式的 psychotherapy research. Contemporary Psychoanalysis, 52(2), 275–312.
程度是由于治疗师或病人的特点的结论。 [Link]
Lundh, L. G. (2015). The person as a focus for research—The contributions of
关键词: 反移情, 治疗师回应问卷, 三元, 集群分析 Windelband, Stern, Allport, Lamiell, and Magnusson. Journal for Person-
Oriented Research, 1(1–2), 15–33. [Link]
Lundh, L. G. (2019). The crisis in psychological science, and the need for a
References person-oriented approach. In J. Valsiner (Ed.), Social philosophy of
science for the social sciences (pp. 203–223). Springer. [Link]
Balint, M. (1955). Friendly expanses; horrid empty spaces. The International
10.1007/978-3-030-33099-6_12
Journal of Psycho-Analysis, 36(4–5), 225–241.
Main, M. (2000). The organized categories of infant, child, and adult
Berg, J., Lundh, L.-G., & Falkenström, F. (2019). Countertransference in
attachment: Flexible vs. inflexible attention under attachment-related
Swedish psychotherapists: testing the factor structure of the Therapist
stress. Journal of the American Psychoanalytic Association, 48(4),
Response Questionnaire. Research in Psychotherapy: Psychopathology,
1055–1096. [Link]
Process and Outcome, 22(1). [Link]
Maslach, C., & Leiter, M. P. (2016). Understanding the burnout experience:
Betan, E., Heim, A. K., Zittel Conklin, C., & Westen, D. (2005). Counter-
Recent research and its implications for psychiatry. World Psychiatry,
transference phenomena and personality pathology in clinical practice: An
15(2), 103–111. [Link]
empirical investigation. The American Journal of Psychiatry, 162(5), 890–
McClure, B. A., & Hodge, R. W. (1987). Measuring countertransference and
898. [Link]
attitude in therapeutic relationships. Psychotherapy: Theory, Research, &
Blum, H. P. (1973). The concept of erotized transference. Journal of the
Practice, 24(3), 325–335. [Link]
American Psychoanalytic Association, 21(1), 61–76. [Link]
Money-Kyrle, R. E. (1956). Normal counter-transference and some of its
.1177/000306517302100104
deviations. The International Journal of Psycho-Analysis, 37(4–5), 360–366.
Bolognini, S. (1994). Transference: Erotised, erotic, loving, affectionate. The
Pope, K. S., Keith-Spiegel, P., & Tabachnick, B. G. (2006). Sexual attraction
International Journal of Psycho-Analysis, 75(1), 73–86.
to clients: The human therapist and the (sometimes) inhuman training
Bonasia, E. (2001). The countertransference: Erotic, erotised and perverse.
system. Training and Education in Professional Psychology, S(2), 96–
The International Journal of Psycho-Analysis, 82(2), 249–262. https://
111. [Link]
[Link]/10.1516/Y9UA-JA7U-3XVJ-RVB4
Rogers, C. R. (1957). The necessary and sufficient conditions of therapeutic
Bornstein, R. F. (2007). Nomothetic psychoanalysis. Psychoanalytic Psy-
personality change. Journal of Consulting Psychology, 21(2), 95–103.
chology, 24(4), 590–602. [Link]
[Link]
Britton, R. (2003). Sex, death, and the superego: Experiences in psycho-
Searles, H. F. (1959). Oedipal love in the countertransference. The Interna-
analysis. Karnac Books.
tional Journal of Psycho-Analysis, 40, 180–190.
Dahl, H.-S. J., Høglend, P., Ulberg, R., Amlo, S., Gabbard, G. O., Perry, J. C.
Tanzilli, A., Colli, A., Del Corno, F., & Lingiardi, V. (2016). Factor
, & Christoph, P. C. (2017). Does therapists’ disengaged feelings influence
the effect of transference work? A study on countertransference. Clinical structure, reliability, and validity of the Therapist Response Questionnaire.
Psychology & Psychotherapy, 24(2), 462–474. [Link] Personality Disorders, 7(2), 147–158. [Link]
cpp.2015 per0000146
 GENERAL PATTERNS IN COUNTERTRANSFERENCE 153

Ulberg, R., Amlo, S., Hersoug, A. G., Dahl, H.-S. J., & Høglend, P. (2014). Westwood, S., Morison, L., Allt, J., & Holmes, N. (2017). Predictors of
The effects of the therapist’s disengaged feelings on the in-session process emotional exhaustion, disengagement and burnout among improving
in psychodynamic psychotherapy. Journal of Clinical Psychology, 70(5), access to psychological therapies (IAPT) practitioners. Journal of Mental
440–451. [Link] Health (Abingdon, England), 26(2), 172–179. [Link]
Varga, A., Torma, B., & Bergman, L. R. (2015). ROPstat: A general statistical 09638237.2016.1276540
package useful for conducting person-oriented analyses. Journal for Person- Windelband, W. (1998). History and natural science. Translated and
Oriented Research, 1(1–2), 87–98. [Link] with an introduction by James Lamiell. Theory and Psychology, 8(1),
Wallerstein, R. S. (2009). What kind of research in psychoanalytic science? 5–22. [Link] (Original work pub-
The International Journal of Psycho-Analysis, 90(1), 109–133. https:// lished 1894)
[Link]/10.1111/j.1745-8315.2008.00107.x Wishart, D. (1987). Clustan user manual. Clustan.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

E-Mail Notification of Your Latest Issue Online!

Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at [Link] and you will
be notified by e-mail when issues of interest to you become available!