Psychology Transference

Abstract

Psychology transference is a type of psychotherapeutic process: the redirection onto a present person, classically the therapist, of feelings and expectations first formed in an important past relationship. Freud named it after Breuer's cathartic treatment of hysteria exposed it clinically, and psychoanalysis made its interpretation the engine of cure while distinguishing it from the working alliance that carries the therapy. A second, experimental tradition recast transference as ordinary social cognition: a mental representation of a significant other, once activated by a resembling new person, is applied to that person, reproducing old inferences, feelings, and self-states. Controlled psychotherapy research has tested whether interpreting transference improves outcome, finding the strongest benefit for patients with the most impaired relationships. This article traces both strands and, through three demonstrations, shows transference as the memory-driven reuse of a relational template.

Keywords: transference, relational self, significant-other representation, working alliance, transference interpretation

What Psychology Transference Is

Transference is the displacement onto a present relationship of affect, expectation, and behaviour that belong, in origin, to an earlier and formative one. In the MeSH definition it is the redirection of feelings and desires — especially those unconsciously retained from childhood — toward a new object, the classic new object being the psychoanalyst. It is filed as a psychotherapeutic process, a within-treatment mechanism, and it has one narrower child term of its own, countertransference, the analyst's answering redirection onto the patient (Freud, 1912).

The phenomenon is easiest to see in its clinical form. A patient comes to feel toward the therapist an intensity — of longing, resentment, suspicion, or devotion — out of proportion to the brief, professional acquaintance that the therapy actually is. The excess is the tell. It marks feeling that has been carried from elsewhere, a template laid down in a relationship with a parent or other early figure and reactivated by the present one. Freud's decisive move was to treat this not as a nuisance interrupting the work but as the work itself: the past made present and therefore, for the first time, observable and open to revision (Breuer & Freud, 1895).

This article follows two traditions that the single word has come to name. The first is clinical and psychoanalytic: the discovery of transference in the treatment of hysteria, its elevation into the instrument of cure, and the eventual controlled tests of whether interpreting it actually helps. The second is experimental and cognitive: the demonstration that transference is a special case of ordinary social cognition — the routine reuse of stored representations of significant others in the perception of new people (Andersen & Chen, 2002).

From Freud's Discovery to a Clinical Concept

The clinical observation came first, in Josef Breuer's treatment of the patient known as Anna O. in the early 1880s. Breuer found that his patient developed toward him an intense personal attachment that had nothing to do with him as a man and everything to do with feelings transposed from her own history; the attachment frightened him and he broke off the treatment. Reporting the case with Freud in Studies on Hysteria, the phenomenon appears as a clinical fact before it has a name — the reproduction, in the doctor-patient relationship, of feeling belonging to an earlier scene (Breuer & Freud, 1895).

Freud named it and, over the following two decades, reversed its meaning. Where Breuer had fled the attachment as an accident of treatment, Freud came to see transference as inevitable, universal, and indispensable. In The Dynamics of Transference he argued that every patient forms a transference because everyone carries templates of loving and relating laid down in early life and imposes them, unconsciously, on each new significant figure. In therapy those templates fasten onto the analyst, and the resulting transference — whether a positive transference of affection and trust or a negative transference of hostility and suspicion — becomes the arena in which the patient's central conflicts are revived in the present and made available for interpretation. In its most concentrated form this reliving takes shape as a transference neurosis, an artificial neurosis enacted toward the analyst that classical technique treated as the very ground of cure. The obstacle became the instrument (Freud, 1912).

Freud gave this reliving a name and a mechanism. In Remembering, Repeating and Working-Through he argued that what a patient cannot consciously remember they instead repeat, reproducing the repressed past not as recollection but as action in the present relationship. Transference is the principal medium of this repetition compulsion: the patient does not recall the old conflict so much as re-enact it toward the analyst. The therapeutic answer is not a single cathartic insight but working-through, the patient's repeated confrontation with the same pattern as it recurs, until its grip on the present loosens (Freud, 1914).

The concept required refinement before it could be worked with cleanly, because not everything a patient feels toward a therapist is transference. Ralph Greenson drew the distinction that organizes modern practice: between the transference proper — the distorted, past-driven, and often neurotic dimension of the bond — and the working alliance, the rational, cooperative relationship in which patient and therapist agree to collaborate on the task. The working alliance is comparatively realistic and undistorted, and it is what allows a patient to tolerate the frustrations of treatment and to examine their own transference rather than merely live it out. Confusing the two, Greenson warned, leads therapists to interpret as neurotic distortion what is in fact the healthy collaboration the therapy depends on (Greenson, 1965).

Types of Psychology Transference

MeSH places transference within the psychotherapeutic processes and gives it one narrower descriptor directly beneath it, distinguishing the two directions the redirection can run between patient and clinician.

The parent-child pairing is deliberately symmetric: transference is the patient's redirection onto the therapist, countertransference the therapist's redirection back onto the patient. The two are not independent — a patient's provocations can call up a therapist's countertransference, and a therapist's disengaged feelings can in turn diminish the benefit the patient draws from having their transference interpreted (Dahl et al., 2017). The MeSH tree is an indexing classification rather than a theory of mechanism: it records that countertransference is catalogued as a kind of transference for retrieval purposes, not that the two are psychologically identical. Countertransference does not yet have its own article on this site, so it is named here without a link.

The Social-Cognitive Reformulation

A second tradition, beginning in the late 1980s, took transference out of the consulting room and into the laboratory, and in doing so changed what kind of thing it is thought to be. Susan Andersen's insight was that transference need not be a special pathology of therapy at all: it could be a routine consequence of how memory stores relationships. Each person accumulates rich mental representations of the significant others in their life, and those representations, like any knowledge structure, can be activated by a resembling cue and then applied to whatever is at hand — including a new person who happens to fit (Andersen & Cole, 1990).

The program tested this with a now-standard method. Participants first describe their own significant others; later, in an ostensibly unrelated study, they meet a new person described with a few features that, for some participants, overlap with their own significant other. The prediction is that the overlap activates the stored representation, which is then applied to the newcomer. It is confirmed repeatedly: participants misremember the new person as having attributes they were never given but that belong to the significant other, they take on the evaluative tone — warm or cold — that the significant other evokes, and their sense of themselves shifts toward the self they are when with that person (Andersen & Chen, 2002).

Three findings sharpen the account. First, the activation can occur without awareness: significant-other cues presented subliminally still trigger the representation and its inferences, so the reader need not consciously notice the resemblance for transference to run (Glassman & Andersen, 1999). Second, transference is not merely a bias in perception but a shaper of behaviour: participants who have a significant-other representation activated actually behave toward the new person in ways that pull the expected responses from them, a behavioural confirmation that can make the transference self-fulfilling (Berk & Andersen, 2000). Third, the phenomenon is anchored in a broader theory of the relational self — the idea that the self is not a single structure but a family of context-bound selves, each linked to a significant other and each summoned when that other's representation is active — which situates transference within the shared, interpersonal construction of meaning (Andersen & Przybylinski, 2018).

Testing Transference in the Clinic

If transference is the instrument of psychoanalytic cure, then interpreting it should improve outcome — a claim that went largely untested for most of a century. The first requirement was to measure transference reliably. Lester Luborsky's Core Conflictual Relationship Theme (CCRT) method supplied it, extracting from a patient's narrated relationship episodes a recurrent pattern in three parts: a wish, an anticipated response from the other, and a resulting response of the self. The recurring theme, appearing across relationships and reappearing toward the therapist, operationalized transference as something that could be scored from transcripts rather than merely intuited (Crits-Christoph, Cooper, & Luborsky, 1988).

With measurement in hand, the effect of interpretation could be tested experimentally. Per Høglend's First Experimental Study of Transference (FEST) randomized patients in dynamic psychotherapy to treatment with or without transference interpretations, holding the therapy otherwise constant. The headline result overturned a clinical assumption. Transference work did not most help the healthiest patients; its sustained benefit was concentrated among patients with more impaired object relations — those with the poorest lifetime history of relationships — for whom examining the here-and-now relationship with the therapist did lasting good (Høglend et al., 2008). Reviewing the wider evidence, Høglend later argued that the exploration of the patient-therapist relationship is a specific and dosable ingredient, beneficial in moderation and for particular patients rather than a universal good to be maximized (Høglend, 2014).

Two further results fill in the picture. The effect of transference work is modulated by the therapist's side of the relationship: when therapists reported disengaged, withdrawn feelings toward a patient, the benefit of transference interpretation was reduced, evidence that countertransference is not merely a curiosity but a moderator of technique (Dahl et al., 2017). And the FEST design has since been extended to adolescents in a randomized trial of transference work with teenagers, carrying the experimental test of the concept into a new developmental population (Ulberg et al., 2021). Meanwhile the concept has spread beyond its psychoanalytic home: transference is now recognized as operating across therapeutic orientations, including nominally non-analytic ones, wherever a patient's relational history colours the alliance with a clinician (Gelso & Bhatia, 2012).

Transference in Motion

The three demonstrations below make the two traditions manipulable. The first builds the social-cognitive account, showing how the overlap between a new person and a stored significant-other representation drives the inferences transferred onto them. The second turns the CCRT method into a working extractor, deriving a recurrent relationship theme from narrated episodes. The third reconstructs the central FEST finding, tracing how the benefit of transference interpretation depends on a patient's quality of object relations.

Significant-other activation: how a stranger inherits a past relationship

A new acquaintance is described by the five features below. Mark the ones that match someone important from your past. The more they overlap, the more the stored representation is activated and applied.

Overlap is 3 of 5 described features (o = 0.60). The account predicts 3.0 of the five unstated attributes are inferred onto the acquaintance, and a warmth of +0.48 is imported — feeling the stranger never earned, carried over from someone they merely resemble.

The activation demonstration makes Andersen's mechanism visible. Selecting which features of a new acquaintance overlap with a stored significant-other representation recomputes the degree to which that representation is activated, and with it the number of unstated attributes inferred, the evaluative tone imported, and the shift in the active self — showing why a stranger who merely resembles someone important can inherit feelings they never earned.

The Core Conflictual Relationship Theme: extracting a transference pattern

Each relationship episode is scored as a wish, an expected response from the other, and a response of the self. Toggle which episodes are in the sample; the most frequent component of each becomes the recurrent theme.

Pervasiveness: 72% of the 18 scored components across 6 episodes match the modal theme. The higher the pervasiveness, the more the same relational template repeats — and the more likely it is to reappear, as transference, toward the therapist.

The CCRT demonstration turns Luborsky's method into a working tool. Adding relationship episodes, each tagged with a wish, an expected response from the other, and a response of the self, extracts the most frequent component of each and assembles the composite theme, with a pervasiveness score that rises as the same pattern recurs — the quantified core of transference as a repeating relational template.

Transference work: who does interpreting it actually help?

Predicted gain: 6.5. Interpretation adds +3.5 at this quality of object relations — a substantial benefit for a more impaired patient, the study's central and counterintuitive result.

The transference-work demonstration reconstructs the FEST interaction. Setting a patient's quality of object relations and toggling transference interpretation on or off traces the predicted gain in psychodynamic functioning, reproducing the study's counterintuitive crossing: interpretation helps the patients with the most impaired relationships most, and adds little for those who are already relationally healthy.

Worked Example

Take the social-cognitive activation the first demonstration draws. Model a significant-other representation as a set of ten features, of which a new acquaintance is explicitly described with a subset; the rest are candidates the perceiver may falsely infer. Let the overlap be the fraction of described features that match the representation, and let representational activation rise linearly with overlap.

Suppose the acquaintance is described with five features, of which three match the significant-other representation. The overlap is o = 3 / 5 = 0.60. Model activation as a = o, so a = 0.60. The stored representation contains five further attributes the acquaintance was never described as having. The social-cognitive account predicts these are inferred — the perceiver goes beyond the information given — in proportion to activation. With inference probability p = a for each unstated attribute, the expected number of falsely inferred (transferred) attributes is 5 × 0.60 = 3.0.

Now compare a control acquaintance who shares only one of five features with the representation: o = 1 / 5 = 0.20, a = 0.20, and the expected transferred attributes fall to 5 × 0.20 = 1.0. The high-overlap acquaintance is predicted to accrue three inferred attributes to the control's one — a threefold increase in transferred content — even though both were described with exactly five real features. The evaluative tone imported scales the same way: if the significant other carries an affect valence of +0.8 on a −1 to +1 scale, the transferred valence is a × 0.8, giving +0.48 for the high-overlap acquaintance versus +0.16 for the control. The arithmetic does not prove the representation was activated, but it makes precise the experiment's signature: matched real information, divergent inferred information, with the divergence tracking overlap rather than anything the new person actually did (Andersen & Cole, 1990).

Discussion

Transference is one of the rare psychoanalytic concepts to have survived translation into experimental psychology, and the translation clarifies what the clinical tradition had described. Freud's claim that everyone imposes templates of early relationships onto present ones is, stripped of its drive-theoretic scaffolding, a claim about memory: that relationships are stored as structured representations and that those representations are reused, automatically and often unconsciously, whenever a new person resembles the person they encode (Andersen & Chen, 2002). The social-cognitive program converted an interpretive claim into a set of predictions that could be, and were, confirmed under controlled conditions.

The two traditions correct each other. The clinical tradition supplied the phenomenon and the stakes — transference as the material of cure — but was slow to test its own central technical claim. The experimental tradition supplied rigor but works with brief laboratory encounters rather than the years-long, high-affect relationships of real therapy. Where they meet is instructive: the FEST trials show that interpreting transference is neither the universal solvent that classical technique assumed nor an inert ritual, but a specific intervention whose value depends on the patient, helping most exactly those whose relational templates are most impaired (Høglend et al., 2008). Greenson's older distinction between transference and the working alliance anticipates this: the part of the relationship that is distorted by the past is the part interpretation targets, while the realistic collaborative part is what makes the interpretation bearable (Greenson, 1965).

For cognitive psychology the value of transference is as a bridge concept. It shows a clinical construct grounded in the same machinery — representation, activation, inference, behavioural confirmation — that governs ordinary person perception, and it shows that machinery producing effects with real emotional and interpersonal force. The person who dislikes a new colleague on sight because the colleague resembles a resented sibling, and then behaves in ways that make the colleague cold, is running the same process the analyst watches in the transference and the experimenter measures in the lab. Transference, on this reading, is not an exotic product of the couch but a standing feature of a mind that remembers relationships and reuses them (Berk & Andersen, 2000).

Current Directions

The most active clinical strand is the experimental dismantling of the assumption that transference interpretation is uniformly good and best reserved for healthier patients. The FEST program established the opposite interaction, and current work is extending its randomized design to new populations and refining its moderators. The trial of transference work with adolescents carries the method into a developmental period when relational templates are still consolidating, testing whether the here-and-now focus that helps impaired adults also helps teenagers whose object relations are not yet fixed (Ulberg et al., 2021). In parallel, the study of countertransference as a measurable moderator — the finding that a therapist's disengaged feelings blunt the benefit of transference work — is turning the analyst's side of the relationship into a variable that trials can track rather than a source of noise to be controlled away (Dahl et al., 2017).

The social-cognitive strand has broadened from transference proper to the relational self and the shared construction of meaning between people. Recent work locates significant-other representations within a wider account of how relationships build and maintain a shared reality — the sense of a common view of the world sustained with close others — and asks how the activation of a relational self shapes not only perception of a new person but the perceiver's own goals, motivations, and self-regulation in that person's presence (Andersen & Przybylinski, 2018). Across both strands the trajectory is the one Freud's successors began: from transference as a singular mystery of psychoanalysis toward transference as a measurable, moderable instance of how memory for relationships governs present social life.

Common Misconceptions

Transference happens only in psychoanalysis.

It is a general feature of social cognition, not a product of the couch. Laboratory studies show significant-other representations being activated and applied to strangers by ordinary participants, and clinical work finds transference operating across therapeutic orientations, including non-analytic ones (Gelso & Bhatia, 2012).

Everything a patient feels toward a therapist is transference.

Greenson distinguished the transference proper from the working alliance — the realistic, cooperative relationship in which patient and therapist collaborate on the task. Treating the healthy alliance as neurotic distortion is a technical error, and the alliance is what makes examining the transference possible (Greenson, 1965).

Transference requires the person to notice the resemblance.

Activation can be entirely unconscious. Significant-other cues presented subliminally still trigger the stored representation and its inferences, so transference runs without the perceiver being aware of what set it off (Glassman & Andersen, 1999).

Interpreting transference helps the healthiest patients most.

Randomized evidence shows the reverse. The sustained benefit of transference interpretation was concentrated among patients with the most impaired object relations, not those with the healthiest relational histories, overturning a long-standing clinical assumption about who tolerates the technique (Høglend et al., 2008).

Glossary

Behavioural confirmation.

The process by which an activated significant-other representation leads a perceiver to act toward a new person in ways that elicit the very responses the representation expects, making the transference self-fulfilling.

Core Conflictual Relationship Theme (CCRT).

Luborsky's method for scoring transference from narrated relationship episodes as a recurrent triad of a wish, an anticipated response from the other, and a response of the self.

Countertransference.

The therapist's own conscious and unconscious emotional reactions to the patient, driven by the therapist's history; the narrower MeSH descriptor filed directly under transference.

Negative transference.

Transference in which hostile, suspicious, or aggressive feelings from a past relationship are redirected onto the therapist.

Object relations.

The internalized patterns of relating to others formed in early life; their quality — a patient's lifetime history of close relationships — moderates the benefit of transference interpretation.

Positive transference.

Transference in which affectionate, trusting, or idealizing feelings from a past relationship are redirected onto the therapist; in excess it can itself become a resistance to the work.

Quality of object relations.

A measurable index of how mature and satisfying a person's characteristic relationships are; in the FEST trials, lower quality predicted greater benefit from transference work.

Relational self.

Andersen and Chen's theory that the self is a family of context-bound selves, each linked to a significant other and summoned when that other's representation is activated.

Significant-other representation.

The structured mental representation of an important person in one's life that, when activated by a resembling cue, is applied to a new person in transference.

Transference interpretation.

The therapeutic technique of drawing the patient's attention to the transference — the here-and-now relationship with the therapist — as it reproduces past relational patterns.

Transference neurosis.

The concentrated re-enactment, within the analytic relationship, of a patient's central conflicts, which classical technique treated as the arena of cure.

Transference.

The redirection onto a present person of feelings, expectations, and patterns first formed in an important past relationship.

Working alliance.

Greenson's term for the rational, cooperative, comparatively undistorted part of the patient-therapist relationship that carries the collaborative work of therapy, distinct from the transference proper.

Working-through.

The repeated analysis of a transference pattern across many instances until its grip loosens; Freud's successor to the idea that a single cathartic insight could cure.

Key Researchers

Susan M. Andersen. Social psychologist at New York University who built the social-cognitive theory of transference, showing experimentally that significant-other representations, once activated, are applied to new people and reproduce old feelings, inferences, and self-states. Faculty page - Wikidata

Josef Breuer. Austrian physician (1842-1925) whose treatment of Anna O. produced the first clinical observation of transference, the intense personal attachment that Freud would later name and theorize. Wikipedia - Wikidata

Sigmund Freud. Founder of psychoanalysis (1856-1939) who named transference and, in The Dynamics of Transference, reframed it from an obstacle into the central instrument of analytic work, arguing that everyone imposes templates of early relationships onto present ones. Wikipedia - Wikidata

Ralph R. Greenson. Psychoanalyst at UCLA (1911-1979) who distinguished the working alliance from the transference neurosis, clarifying which part of the patient-therapist bond does the therapeutic work. Wikipedia - Wikidata

Per Høglend. Psychiatrist and psychotherapy researcher at the University of Oslo who led the First Experimental Study of Transference, the landmark randomized trials testing whether transference interpretation improves dynamic-psychotherapy outcome. Faculty page - Wikidata

Lester Luborsky. Psychotherapy researcher at the University of Pennsylvania (1920-2009) who created the Core Conflictual Relationship Theme method, the first reliable empirical operationalization of transference from psychotherapy transcripts. Wikipedia - Wikidata

Randi Ulberg. Psychiatrist and researcher at the University of Oslo and Diakonhjemmet Hospital who led the FEST-IT randomized study extending the experimental test of transference work to adolescents. ORCID - Faculty page

Frequently Asked Questions

What is transference in psychology?
Transference is the redirection onto a present person, classically the therapist, of feelings, expectations, and relational patterns first formed in an important past relationship. In MeSH it is defined as the redirection of feelings and desires, especially those unconsciously retained from childhood, toward a new object, and it is filed as a psychotherapeutic process (Freud, 1912).

Where does the concept come from?
It emerged from Breuer's treatment of Anna O., whose intense attachment to her physician appeared in Breuer and Freud's Studies on Hysteria as a clinical fact before it had a name. Freud named it and, in The Dynamics of Transference, made its interpretation the central instrument of psychoanalysis (Breuer & Freud, 1895).

How is transference different from the working alliance?
Greenson distinguished the transference proper, the distorted, past-driven dimension of the bond, from the working alliance: the realistic and cooperative relationship in which patient and therapist collaborate on the task. The alliance is comparatively undistorted and is what allows a patient to tolerate treatment and examine their own transference (Greenson, 1965).

Is transference only a psychoanalytic idea?
No. Experimental work shows it is a routine feature of social cognition: a stored representation of a significant other, activated by a resembling new person, is applied to that person by ordinary participants in the laboratory. Clinically, transference is recognized across therapeutic orientations, including non-analytic ones (Andersen & Chen, 2002).

Can transference happen without a person noticing?
Yes. Significant-other cues presented subliminally still activate the stored representation and its inferences, so a person can transfer feelings onto a new acquaintance without ever consciously registering the resemblance that set it off (Glassman & Andersen, 1999).

Does interpreting transference actually help patients?
It depends on the patient. The First Experimental Study of Transference found that the sustained benefit of transference interpretation was concentrated among patients with more impaired object relations, not the healthiest ones, overturning the assumption that only relationally healthy patients tolerate the technique (Høglend et al., 2008).

What is the CCRT?
The Core Conflictual Relationship Theme is Luborsky's method for measuring transference from a patient's narrated relationship episodes. It extracts a recurrent triad (a wish, an anticipated response from the other, and a response of the self) that repeats across relationships and reappears toward the therapist (Crits-Christoph, Cooper, & Luborsky, 1988).

How is transference related to countertransference?
Countertransference is the therapist's own redirection onto the patient, the mirror of the patient's transference and the one narrower descriptor MeSH files beneath it. It is not merely a curiosity: when therapists report disengaged feelings toward a patient, the benefit of transference interpretation is measurably reduced (Dahl et al., 2017).

References