Collusion Revisited: A Narrative Review of Dyadic Collusions

The concept of collusion is defined as an unconscious, unresolved issue that ties together two or more people, who are interlocked in a defensive maneuver that allows them to externalize the issue and avoid it at the intrapsychic level (see the working definition and detailed description provided below). Collusion has crossed the boundaries of psychotherapeutic schools (it is used in psychoanalytic and systemic therapy) and entered the medical field (Atkinson & McNamara, 2017), and it plays a role outside the health care system (e.g., in prisons (Vanderstukken et al., 2015), schools (Schruijer, 2013), industry (Petriglieri & Petriglieri, 2020), and literary analyses (Kaibr & Guo, 2018; Stiefel et al., 2017, 2018, 2019).

This critical narrative review relies on a corpus of articles that we selected to produce the first manuscript addressing collusion in a palliative care setting (Stiefel et al., 2017). The corpus was increased through the constant screening of these and subsequently retrieved articles to obtain additional references on collusion. Since descriptions of collusion appeared in articles on projective identification or enactment, we also screened and explored this literature. We reviewed more than 250 English, French, and German articles from the psychoanalytic, systemic, social psychology, and system psychodynamics literature. The consultation of approximately 30 books dealing with collusion or aspects relevant to collusion, such as Henry Dicks “Marital Tensions” (Dicks, 1967), Jürg Willi’s “Couples in Collusion” (German edition: “Die Zweierbeziehung”) (Willi, 1975) or Cassorla’s “The Psychoanalyst, the Theatre of Dreams and the Clinic of Enactment” (Cassorla, 2018), completed this study of the literature. Our overriding objective was to provide a comprehensive perspective on collusion based on the literature that was also informed by our experiences as liaison psychiatrists, psychotherapists, and supervisors (FS and MS) and as a social scientist (CB), having been embedded for more than a decade in the same psychiatric liaison service; together, we have started to conduct research on collusion. Some of the clinical illustrations are thus derived from our clinical activity, which might be less familiar to the readers. A critical reflective stance led us to question some ideas and propositions contained in the literature and to add new thoughts, which contributed to the tasks of situating, defining, delineating, and revisiting collusion.

The manuscript is divided into three parts. The first part situates dyadic collusion, the second part comprehensively and critically reviews multiple clinical facets of this notion, and the third part classifies and comments on different types of dyadic collusions.

Situating Collusion

The term collusion refers to different phenomena. According to a dictionary definition, collusion is commonly understood—as in the legal field or in industry—as a secret agreement and cooperation for illegal purposes between people who share an unvoiced intention to defraud something (Ayres, 1987). In the management literature, collusion is considered to be the result of complementary relationships (e.g., the need of an autocratic boss to have subordinates who are willing to collusively obey and vice versa) (Harvey, 1974) or of produced and reproduced norms and practices (Cooper et al., 2021). In medicine, collusions are also described as tacit agreements, for example, between physicians and families who withhold diagnostic information from patients (Chaturvedi et al., 2009).

In psychology, collusion is polysemic. In social psychology, for example, collusion describes processes in which a group fails to act in line with the preferences or opinions of its individual members since premature consensus hampers the explicit expression of members’ points of views (Schruijer, 2013).

In this article, we focus on the psychoanalytic definition of collusion. To ensure a common understanding, we first provide a detailed working definition of collusion, which is enriched throughout the manuscript by conceptual add-ons and clinical examples, with the aim of distinguishing collusion from other transference–countertransference experiences. To the best of our knowledge, this endeavor to define, situate, delineate, and revisit collusion has not been undertaken to date. We have added the word “critical” to qualify our narrative review since we try to question, challenge, and reflect (Ng et al., 2019) on the ways in which scholars have treated the concept of collusion. The knowledge gap we thus fill with our review is as follows: we provide a working definition of collusion and distinguish collusion from other transference–countertransference interactions, systematically address its clinical facets and provide a typology of collusion, question some conceptual aspects, clarify the semantic field, and critically discuss the extant literature. This overview thus provides a framework that can be used by clinicians, supervisors, and researchers.

A Working Definition of Collusion

Collusion refers to a specific relational mode that emerges between two or more participants. At the core of collusions are unresolved issues, which are unconsciously shared by the participants and cause the colluders to become interlocked with one another in a defensive, interpersonal maneuver (externalization). The defensive loop in which the unresolved issue is put into play allows colluders to avoid it at the intrapsychic level (Cassorla, 2001; Dicks, 1967; Stiefel et al., 2017; Willi, 1975).

Unresolved issues, which are discussed in the literature under different names such as traumatic issues, blind spots, illusions, or bastions, all of which refer to the notion of that which is repressed and inaccessible to insight, are related to control, intimacy, loss, dependency, and exigency (Stiefel et al., 2017). Such issues, which are frequently encountered in medicine, psychiatry, and psychotherapy, are part of human existence. However, when they are unresolved, they can fuel collusions. The terms “symmetrical” and “complementary” collusion indicate that the colluder can adopt either the same stance toward the unresolved issue (e.g., both being dependent individuals) or an opposite or polarized stance (e.g., one person is dependent while the other adopts a rigid, demonstrative, and defensive position of independence).

Collusions are triggered by verbal exchange, nonverbal elements such as unexpressed emotions or changes in the setting, and nondiscursive factors such as diagnostic or therapeutic procedures in medicine or life events linked to the unresolved issue. Triggers may cause the relationship to collapse into symbiotic stagnation or burst into pieces. Collusions can thus have major relational consequences or negative impacts on clinical decision-making. Collusions occur in psychiatry, psychotherapy, and medicine (Atkinson & McNamara, 2017; Byrne et al., 2002) as well as nonclinical settings such as schools or prisons (Vanderstukken et al., 2015), couples (Zeitner, 2003), families (Yahav & Sharlin, 2002), and natural groups (Goldblatt et al., 2015).

The factor that distinguishes collusion from other transference–countertransference reactions is the relational mode between colluders, which we propose to call “collusive resonance”. Resonance between people can result in growth through cross-fertilization, increased mutual understanding, and coevolution (Willi, 1975). Collusive resonance, which is based on defensive needs, freezes the relationship, stops evolution, and leads either to symbiosis or rupture. We have observed that collusive resonance affects actions or transactions more than other transference–countertransference experiences since it involves the same unresolved issues. In a previous research project, in which we delineated collusions from other transference–countertransference interactions, we obtained results that seem to confirm this hypothesis (Deliyanidis et al., 2023) (i.e., results indicating that collusion, unlike other transference–countertransference interactions, leads to deviation from good medical practice).

Countertransference and Collusion

Conceptual interrelations exist between collusion and countertransference, especially between collusion and transference–countertransference experiences. However, collusion is not equivalent to transference–countertransference because noncollusive transference–countertransference reactions are not provoked by a shared unresolved issue (Stiefel et al., 2017; Willi, 1984). For example, a patient’s anxiety (unresolved issue related to separation), which is expressed by the patient clinging to the therapist, may provoke countertransference anger, which is motivated by the therapist’s feelings of being invaded (an unresolved issue related to intimacy). Moreover, transference and countertransference are limited to therapeutic relationships, but collusion also occurs in nonclinical situations such as couples and natural groups.

We regularly face doubts regarding the specificity of collusion. The critique is based on the following arguments: (i) collusions are merely ordinary transference–countertransference phenomena; (ii) unresolved issues in patients and analysts were recognized decades ago, and the concept of collusion does not offer anything new; and (iii) collusion is equivalent to projective identification or (iv) enactment. We disagree. We maintain that (i) noncollusive transference–countertransference interactions have different functions and clinical qualities: the protagonists are not interlocked in an activated defensive loop, which makes it easier for them to free themselves from their reactions and to start thinking again. For colluders, even if their situation may feel very distressing, the relational bond is entertained by the protagonists, who both “add fuel to the fire.” The same claim holds true for (ii) the unresolved issue: it makes a difference whether an unresolved issue is shared or not. The blind spots of the colluders pertain to the same issue, which diminishes the likelihood of the colluders being able to see what ties them together. Some authors reduce collusion to a manifestation of (iii) projective identification, and it is indeed observed that the two phenomena are frequently associated. However, projective identification does not automatically lead to collusion, and not all collusions are based on projective identification. Other defenses may also be at work (see section below). Finally, (iv) enactment is a nonspecific reaction, which becomes manifest in a wide range of phenomena and motivations that aim to diminish intrapsychic pressure through action.

Collusion and Projective Identification

From an epistemological point of view, the identification of collusion is preceded by the recognition of interactional dimensions in psychoanalysis, unconscious and bidirectional communication between patients and therapists, and interpersonal defense mechanisms, such as projective identification.

Collusions are often mentioned in the context of projective identification (Cassorla, 2001), with projective identification fueling collusive interlocking. Colluders thereby avoid intrapsychic conflict through an interpersonal arrangement but remain imprisoned in a deadlock with their old object relations, repeating past experiences (Dicks, 1963).

Defense Mechanisms and Collusion

The issue of which defenses are at work in collusion remains largely unaddressed in the literature, with the exception of habitual references to crossed projective identification. We believe that other defense mechanisms may be involved in collusion. Denial—for example, denial of a deadly disease—that is shared by a patient and his partner alongside an unresolved issue related to separation may lead to collusion, which is entertained through mutually “reassuring” discourses. The same claim holds true for idealization, which can become manifest in the form of narcissistic collusion, as described by Willi (1975). In symmetrical narcissistic collusions, the shared unresolved issue of low self-esteem is evacuated by means of mutual idealization. In complementary narcissistic collusion, the idealized colluder feels superior and perceives the other as inferior. In a recent study, Kleiner-Paz and Nasim (Kleiner-Paz & Nasim, 2021) argue that dissociation might operate in the context of collusions over intimacy. A special form of collusion is the production of a proxy. This type of collusive interaction is illustrated by the “archetypical” situation of the absent parents: two children are alone at home, and the younger sibling serves as a container for the elder sibling’s anxiety. The older sibling induces fears in the younger sibling and then adopts an appeasing attitude toward him (Wangh, 1962).

Finally, secondary defenses such as rationalization or reaction formation are frequently present in collusion. These “manifest” defenses are in the service of the underlying primary defenses. Displacement may also occur, for example, when emotions induced by collusion are directed toward another person, such as another patient or collaborator (Grinberg, 1962).

The Clinics of Collusion

In this second part, we utilize examples drawn from our supervisory activity and the literature to illustrate the clinical facets of collusion and flesh out the conceptual elements provided to date.

Triggers

Triggers related to unresolved issues increase collusive resonance. In a psychotherapeutic setting, triggers can emerge from the therapeutic frame (e.g., a break in sessions due to holidays can activate unresolved issues related to separation) (Hilty, 2020), from the contents addressed in therapy (e.g., the investigation of traumatic events can activate unresolved issues related to intimacy) (Fox & Carey, 1999) or from nonverbal elements (attitudes, gestures, tone of voice, facial expression) as well as from symptoms, habits or stereotyped behaviors (Kestenberg, 1972). In the medical setting, triggers can be health issues, diagnostic procedures, bodily symptoms, or the delivery of bad news, which can provoke the eruption of unresolved issues, such as intimacy, loss, and self-worth (Stiefel et al., 2017).

In couples, life events can be triggers: the occurrence of a disease in one partner may, for example, lead to unresolved issues concerning dependency (Delvey & Hopkins, 1982); past life events have also been described as triggering collusion in psychotherapy (Welldon & Hacker, 2012).

Karlsson (2004) described collusion in the context of separation triggered by a dream. The psychotic patient had been receiving treatment for years and showed signs of improvement despite massive resistance. This success required considerable effort from the therapist, who had to endure repeated attacks from the patient with regard to their relationship. The patient announced that he would move to another town and therefore had to terminate therapy. In the last session, he reported a dream, which he immediately qualified as completely insignificant. The therapist insisted on focusing on the dream and even on writing it down. This insistence provoked an intense reaction in the patient, who felt threatened by intrusion. We understand this interaction to be an instance of collusion. Both the patient and the therapist, who had invested in the patient over the years despite his repeated attacks, were affected by the imminent separation. Otherwise, the therapist, who knew the patient very well, would simply have accepted the dream as a “departure gift” (our suggestion). In this context, we would like to underline the fact that the information concerning the unconscious dynamics at work in the description of collusion are rather scarce, even when the authors are psychoanalysts (Cassorla, 2001; Karlsson, 2004). This scarcity may be the result of privacy protection.

An instance of collusion related to attachment is derived from a psychotherapy session conducted by one of the authors. The patient peppered the therapist with questions. This situation provoked a growing irritation in the therapist, who started to distance himself from the patient, with the result that the patient accused him of being “cold.” During the session, the therapist recognized that the patient’s multiple questions were an expression of her clinging tendencies (anxious-preoccupied attachment) and that his irritation was a defensive reaction due to his own attachment difficulties (dismissive avoidance). Indeed, the patient’s development was marked by a conflictual relationship with her mother and a rather absent father, resulting in attachment difficulties and a functional bowel disorder. The therapist’s development was marked by intense and chronic intergenerational conflict, leading to attachment difficulties and panic attacks in early adulthood, which were resolved after psychoanalysis.

Modalities of Collusive Bonds

To the best of our knowledge, Willi is the only author to propose a meta-psychology by classifying complementary collusions as narcissistic, oral, anal-sadistic, phallic, or narcissistic (Willi, 1975, 1984). For example, in oral collusions, the unresolved issue concerns “nurturing.” The so-called progressive caregiver represses oral needs and vicariously experiences them through the receiver, who occupies a regressive position. Jacobs (1986) reports a case of such symmetrical oral collusion: the patient fed the therapist with abundant transference material, while the therapist in turn fed the patient with a transference interpretation. The topic threatening both the patient and therapist who engaged in this “stimulating” psychotherapy, however the dying of the patient’s husband remained unaddressed due to the collusion.

One may ask whether the psychic structure determines the modalities of collusive interactions. We have no arguments for such a hypothesis; Karlsson’s (2004) previously mentioned collusion with a psychotic patient did not indicate any specificities. In the context of psychosis, one can question whether the so-called folie à deux or folie en famille could be considered to represent collusion. We disagree, given the observation that delusions often persist after the separation of the protagonists (Arnone et al., 2006).

Manifestations of Collusion

Collusions become manifest through thoughts, attitudes, behaviors, the predominance of interactional dimensions in the encounter, deviation from good clinical practice, or intense emotions (Nos, 2014). By definition, collusion can only be recognized in retrospect, for example, after enactments or when therapists feel estranged by their own reactions (Cassorla, 2001). Because enactments often provide a clue regarding collusion, we briefly discuss collusive enactments. Enactments, which are equated by some authors with collusion (Severo et al., 2018), occur when a therapist responds in a manner that reflects the influence of the patient’s projection (Gabbard, 2020). Enactments, which are conceptualized as jointly created nonverbal actualizations of intrapsychic configurations, may be normal and resolved through thought and interpretation. However, they are collusive when they arise in response to a shared, unresolved issue (Cassorla, 2001), as described in the treatment of a patient with an eating disorder (Gubb, 2014). The therapist’s constant hunger and almost hallucinatory visualization of a pizza were not, as initially assumed, a response to the patient’s anorexic behavior. The therapist finally recognized that the collusion was related to a shared and unresolved issue concerning competitiveness, which she understood when she was eating pizza, a dish that she considered to be unhealthy and usually avoided. Unable to resist the temptation of unhealthy food and feeling guilty, the author realized that she was in competition with her patient, who was able to resist eating.

In couples, collusions may manifest in the form of a complete role reversal between partners after an event that disturbs the relational equilibrium, shared acting out due to the same unresolved issue (Godfrind-Haber & Haber, 2002), a blurred distinction between the perception and real existence of the other, deception when the partner fails to behave in the attributed way, ritualized behavior that is incomprehensible to a third party, or statements by partners that their sole problem is the fact that the other exists (Willi, 1975).

The following supervision illustrates the manifestations of complementary oral collusion: A young nurse had just started to work in palliative care and requested supervision because she felt exhausted and feared that she would have to ask for sick leave. She presented the case of an elderly patient with advanced lung cancer, with whom her relationship was initially harmonious. The patient had high expectations regarding medical care, and the nurse was proud to meet those expectations. However, the increasing demands of the patient caused their relationship to deteriorate. The nurse began to find excuses to avoid the patient. In response, the patient increased his demands, and he finally started to criticize her. This criticism provoked intense irritation and a great deal of anxiety on the part of the nurse, given her exigencies toward herself. During supervision, the supervisee realized that her high expectations and idealized identity as an unconditionally devoted (and nurturing) nurse contributed to the dynamics of her relationship with this patient.

Primary and Secondary Gains and Consequences of Collusive Defense

The primary gain obtained through the collusive interpersonal maneuver is the avoidance of an unresolved issue at the intrapsychic level (Dicks, 1963; Willi, 1975, 1984). Possible secondary gains (whether in the context of natural couples or patient-therapist couples) include gratifications associated with the attributed roles, vicarious participation, emotional discharge, prevention of separation, protection from painful issues and control over the object. However, collusion has certain consequences, including the distortion of reality, the repression of parts of the self, and the loss of self-object differentiation (Loewald, 1986).

Effects of Collusion on the Therapeutic Process and Coevolution in Couples

Some authors argue that collusion may strengthen the therapeutic alliance at the beginning of treatment (Godfrind-Haber & Haber, 2002). Such collusions, which are called “necessary collusion” (Cassorla, 2001) or “therapeutic collusions” (Karlsson, 2004), are thought to allow patients and therapists to avoid disillusion that appears to be too rapid. Notably, traumatized patients may remain unable to access trauma (Cassorla, 2018). A clinician who unconsciously adopts a prudent attitude allows the trauma to be addressed only after confidence has been established (Fox & Carey, 1999). However, if the prudence is due to the therapist’s own unresolved trauma, the therapist may ignore clues from the patient that indicate readiness to address the trauma (Fox & Carey, 1999). In this context, it is not appropriate to talk of “necessary or therapeutic collusions.”

Additionally, in psychoanalysis, collusion may remain unrecognized and have negative effects: it can limit reverie (Ogden, 2021), lead to therapeutic ruptures or immobility, imprison the therapist, impede creativity, or break the barriers between the conscious and the unconscious (Cassorla, 2018; Civitarese, 2021). A frequent collusion in the analytic setting is reported by the Barangers, with psychoanalysts being flattered to be viewed as idealized omnipotent figures; in such cases, analysis fails (Baranger & Baranger, 2008). Moreover, collusions may be the origin of abuse in therapy (Teitelbaum, 1991). However, the effects of collusion may not always be dramatic; some authors consider therapists in collusion to be able to continue to exercise a containing function (Cassorla, 2018).

Dicks claims that collusion in couples may be an attempt to overcome unresolved issues and an effort to self-heal by reappropriating lost aspects of the self in the relationship with the other (Dicks, 1967). We agree that some collusions in couples can be attempts at self-healing; however, successful self-healing occurs only through coevolution (cross-fertilization and mutual integration of the complementary characteristics of the partners).

A discussion of collusions involving individuals and groups (e.g., scapegoating) and collusions both within and between groups is beyond the scope of this manuscript and will be described elsewhere.

Facilitating and Maintaining the Factors Associated with Collusion

Factors unrelated to unresolved issues may facilitate or maintain collusions. Institutional rules, for example, can facilitate and maintain collusion; a discussion on how the larger context (e.g., dominant discourses) may facilitate and maintain collusion will also be addressed elsewhere. Role responsiveness or behavioral responsiveness, a primarily unconscious tendency to comply with the expectations of the other (Sandler, 1976), is part of a therapeutic attitude but can, when it becomes excessive, facilitate collusions, for example, a collusion between a “voyeuristic” therapist and an “exhibitionistic” patient (Wood, 2014). Some characteristics of the therapist (hypertrophic ego ideal) or of the patient, such as perversion (Wood, 2014) or suicidality (Nivoli et al., 2014), may facilitate collusions since they diminish the capacity to mentalize.

In couples and families, life events facilitate, maintain, and intensify collusions. Examples include illness and unemployment (Willi, 1975) or specific challenges such as adolescence and senescence (Zinner & Shapiro, 1972).

The Ending of Collusions

Collusion becomes manifest along a spectrum of collusive resonance, which depends on the intensity of the relational dynamics at play, the power of the unresolved issue at stake, the amount of primary and secondary gains to be made, and the severity and type of the collusion’s side effects.

Since collusions have defensive functions, which might fluctuate, they may end naturally (De Beà, 1989). For example, in the medical setting, amelioration of the patient’s condition and the associated decrease in defensive needs may cause collusion to terminate.

In therapy, the clinician can, as with projective identification, address the unresolved issue and the associated interpersonal dynamic or simply contain the situation. The issue of whether a therapist should also acknowledge his contribution to the collusive episode is somehow different than the possible self-disclosure of countertransference, which rarely seems to be beneficial for the patient. In collusion, acknowledging one’s own entanglement may be therapeutic and, to a certain degree, an ethical command. A possible wording for such an acknowledgment might be as follows: “We seem to have gotten entangled in an issue over…”. Such a stance does not imply sharing or detailing the clinician’s own unresolved issue but takes into account the fact that collusion is not just another problem of the patient.

Some authors claim that the very fact that a therapist considers collusion is an indicator of the readiness of the patient to work through it (Cassorla, 2018).

In couples, collusion usually persists, but its intensity may vary, for example, according to life events or challenges throughout the life cycle (Willi, 1975). In couple therapy, collusion can be addressed, or the intervention can focus on the modification of the relational dynamic, as in the case of Bagarozzi (2011), who attributed to a husband the task of serving as a coach for assertiveness training with regard to his dependent wife and thereby attenuated oral collusion.

Collusion and Setting

Collusions are also determined by the setting. In a psychoanalytic setting, which encourages regression in the patient—albeit to a lesser degree than in the analyst (Baranger & Baranger, 2008)—one can assume that collusion is inevitable, since the analyst must enter into communication with the patient, also on an unconscious level. On the other hand, in somatic medicine, collusions have the potential to harm clinician-patient relationships and even impair medical judgment (Atkinson & McNamara, 2017; Stiefel et al., 2017). Therefore, regular supervision can be highly beneficial with regard to identifying and preventing the formation of collusions.

Finally, in the couple therapy setting, collusion, especially in cases of complementary collusive bonds, may be more easily identified. The main reason is that mating, unlike the clinician-patient relationship, is usually based on free choice and that distressing relationships that nevertheless endure draw attention to a possible collusive bond.

The potential harm to the patient‒physician relationship but also to clinical judgment (e.g., oncologists’ therapeutic obstinacy in collusions pertaining to separation anxiety), motivates us to make use of the concept of collusion in the medical setting. This situation raises the question of what elements are sufficient or necessary to assume that collusion is at work. We maintain that a shared and unresolved issue (i), a defensive loop between colluders (ii), and intrapsychic avoidance through externalization (iii) should be demonstrated.

Types of Collusion: A Clarification of the Semantic Field

In this final part of the manuscript, we discuss different types of dyadic collusions and clarify the associated semantic field.

In partial collusions (Willi, 1975), the collusive character of the relationship is recognized, but its origins are defensively intellectualized, thus limiting insight and change. Analogous to objective countertransference (Winnicott, 1949), objective collusions are considered to inevitably occur in patients with severe psychopathology (Teitelbaum, 1991). However, from our point of view, to maintain specificity, the notion of collusion should be restricted to situations in which the therapist shares an unresolved issue with the patient. This situation is not the case with regard to so-called objective collusions. Moreover, we think that this notion is problematic since it delegates all the responsibility to the patient (“everybody would react in this way to this patient”).

Collusion has also been described as an element of empathy (Peabody & Gelso, 1982) based on the assumption that projective identification, when it is not excessive, is part of interpersonal communication and sharing one’s experiences (De Beà, 1989). We believe, as do others, that the notion of empathic collusion is erroneous since empathy does not require a shared unresolved issue and is not defensive. The boundaries among empathy, identification, projective identification, and collusion may at times be difficult to draw, as illustrated by Tansey (1994), who noted that to contact one’s own past pain is not the aim of therapy but might be a byproduct of that process.

One might wonder whether “nonpathological” collusions exist, for example, in human cooperation or as normal responses to universal issues that cannot be “resolved”, such as separation (and death). We agree that cooperation in a complementary mode exists among couples, especially after many years of partnership. The decisive criterion for distinguishing cooperation from collusion is the defensive function of collusion, which serves intrapsychic rather than interpersonal needs, leading to stagnation rather than to coevolution (Willi, 1985). With regard to unresolvable issues, a collusive attitude toward existential issues, such as death, evacuates the threatening issue from the intrapsychic space, whereas a noncollusive attitude evolves and confronts reality without excluding suffering and distress.

By distinguishing between acute and chronic collusions, it can be suggested that acute collusions have a communicative role, while chronic collusions impede the therapeutic process (Cassorla, 2018). We agree that acute collusions, if they are worked through, can contribute to the therapeutic process; however, if they are unrecognized, they may cause harm (see Sect. 2.5).

Children who are subjected to long-lasting projective identification may start to behave collusively. Such collusion may induce structural changes and thus be considered “developmental collusions” (Kestenberg, 1972).

Conclusions

Collusions, when identified, understood, and worked through in psychotherapy, can unlock symbiotization, repair ruptures and dissolve stagnation with regard to therapeutic progress. In couple therapy, the resolution of collusions unfreezes relationships, contributes to the acceptance and reinternalization of projected parts, and helps partners to increase their autonomy and their self-coherence. In supervision, attention to collusion can illustrate transference-countertransference, raise awareness of the clinician’s own contribution to therapeutic impasses and remind supervisors of the possible occurrence of parallel processes. Finally, collusion is a concept that bridges psychotherapeutic currents and, more generally, is a boundary object that unites clinicians with patients, the professional with the private and various fields of research with one another, ranging from psychotherapy, social psychology, system psychodynamics and institutional analysis to psychiatry and medicine.