Psychotherapeutic Processes

Abstract

Psychotherapeutic processes, which MeSH classifies under psychotherapy, are the change-producing events that unfold within treatment rather than the techniques that name a school. This article follows the evidence that these shared relational processes, above all the working alliance, carry much of therapy's effect regardless of orientation. It sets out the common-factors tradition, the tripartite structure of the alliance and its modest but robust link to outcome, the rupture-and-repair cycle, and the harder question of mechanism, whether a strong alliance causes improvement or partly reflects it. Three interactive demonstrations let the reader read a correlation as a change in recovery odds, trace an alliance through rupture and repair, and see how a tripartite alliance is limited by its weakest bond.

Keywords: working alliance, common factors, alliance rupture

Psychotherapeutic processes are the within-treatment events through which psychotherapy is thought to work: the relationship between client and therapist, the shared goals and tasks of the work, moments of strain and their repair, and the client's own engagement, expectation, and emotional processing. They are distinguished from the branded procedures that define a school, cognitive restructuring, free association, systematic desensitisation, and the like, because a central finding of psychotherapy research is that treatments built on very different procedures tend to produce broadly similar outcomes, which turns attention from what is done to a client toward what happens between the two people in the room (Wampold, 2015). In the Medical Subject Headings vocabulary the topic is catalogued beneath psychotherapy as the processes internal to it, and the study of those processes, how change actually comes about rather than merely whether it does, is a field in its own right (Greenberg, 1986). The sections below set out the common factors, the working alliance and its structure, the recognised subtypes of process, the rupture-and-repair cycle, the debate over mechanism, and the roles of empathy and expectation.

The Common Factors

The idea that different therapies share the ingredients that actually produce change is older than most of the techniques it explains. Saul Rosenzweig first argued in 1936 that the diverse methods of psychotherapy owe their common success to implicit factors they share rather than to the theories that distinguish them, and he borrowed a line from Alice in Wonderland, the Dodo's pronouncement that everybody has won and all must have prizes, to name the suspicion that no one school outperforms the rest (Rosenzweig, 1936). Four decades later Lester Luborsky and colleagues gave the conjecture its empirical form, reviewing the comparative-outcome studies and concluding that bona fide treatments produced largely equivalent results, so that the Dodo bird verdict, as the finding came to be known, became the central puzzle that the common-factors tradition exists to explain (Luborsky et al., 1975). Carl Rogers proposed that a small set of relational conditions, a therapist's genuineness, unconditional positive regard, and accurate empathic understanding, were not merely helpful adjuncts but the necessary and sufficient conditions for therapeutic personality change, a claim that placed the relationship, rather than any technical procedure, at the causal centre of therapy (Rogers, 1957). Later reviewers, confronted with the repeated finding that bona fide therapies rarely differ much in outcome, formalised this into the common-factors position: that factors shared across treatments, the alliance, empathy, expectation, and the provision of a coherent rationale and a set of actions consistent with it, account for a large share of the benefit, while the specific ingredients unique to each brand account for rather little (Lambert & Barley, 2001). Bruce Wampold's contextual model draws the argument together, holding that psychotherapy works through a genuine relationship, the client's expectation that treatment will help, and engagement in the specific therapeutic actions, so that technique matters but chiefly as the vehicle through which these common pathways operate (Wampold, 2015). The position is not that technique is inert; it is that the relational and expectational processes are the larger and more reliably measured part of the effect. A meta-analytic review of the question concluded that common factors are strongly associated with outcome and plausibly mediate a substantial portion of it, while cautioning that most of the evidence is correlational and cannot by itself establish that these factors are the active cause (Cuijpers et al., 2019).

The Working Alliance

Of all the common factors, the one that has been measured most and defined most sharply is the working alliance. Edward Bordin recast what psychoanalysis had called the therapeutic relationship into a pantheoretical construct with three components: the emotional bond between client and therapist, their agreement on the goals of treatment, and their agreement on its tasks, the activities each session is made of (Bordin, 1979). Bordin's move was to make the alliance apply to every therapy, not just the psychoanalytic one in which the concept arose, and to make it something two people build collaboratively rather than a transference the client brings ready-made. The construct proved measurable, and the first meta-analysis to pool the resulting studies found a consistent positive relation between the quality of the alliance and the outcome of therapy across treatment types and measures, establishing the alliance as a reliable, if not enormous, correlate of who gets better (Horvath & Symonds, 1991). Three decades of accumulation have not overturned that first result so much as sharpened it. Christoph Flückiger and colleagues, synthesising more than three hundred studies, put the alliance-outcome correlation at roughly r = 0.28, a value that is modest in size but among the most robust and replicated findings in all of psychotherapy research, holding across disorders, treatment approaches, and ways of measuring the alliance (Flückiger et al., 2018). Figure 1 shows how a correlation of that magnitude, small as it looks, translates into a meaningful difference in the odds of improvement, and the demonstration that follows lets the reader vary it.

Types of Psychotherapeutic Processes

Beyond being a subject in its own right, Psychotherapeutic Processes is a formal category in the National Library of Medicine's Medical Subject Headings, which places it at tree position F04.754.720, beneath psychotherapy, and hangs its recognised narrower kinds beneath it. These subtypes are a classification built to index the literature, not a claim about the mind's natural joints; one of them, the therapeutic alliance, is treated at length in its own article. Table 1 lists the direct children of the descriptor.

Two cautions keep this taxonomy in its place. It is a classification for indexing, built to organise the literature, not a theory asserting these subtypes are mutually exclusive or an exhaustive set of natural kinds; abreaction, free association, the alliance, and transference plainly overlap and co-occur within a single session. And a MeSH subtype is a narrower topic, not a component mechanism: listing transference beneath Psychotherapeutic Processes locates it in an index and says nothing, on its own, about whether it is among the processes that actually drive change.

Rupture and Repair

A strong alliance is not a matter of avoiding difficulty; alliances strain, and how a strain is handled may matter more than whether it occurs. Jeremy Safran and colleagues defined an alliance rupture as a deterioration in the collaborative relationship, a breakdown in the bond or a disagreement over tasks or goals, ranging from a barely perceptible withdrawal to an open confrontation, and argued that such moments, far from being failures to be avoided, are opportunities: the collaborative work of noticing a rupture and repairing it can itself be a powerful agent of change, giving the client a corrective experience of a relationship that survives conflict (Safran et al., 2011). The claim is testable, and the evidence supports it. Catherine Eubanks and colleagues, meta-analysing studies that measured rupture and its resolution, found that the presence of repaired ruptures was associated with better treatment outcomes, so that a course of therapy marked by ruptures that are subsequently mended tends to end better than one in which strains go unaddressed (Eubanks et al., 2018). The finding reframes conflict in therapy: a rupture is a fork, and the repair is the therapeutic process. The demonstration below traces an alliance session by session, letting the reader introduce a rupture of variable depth and either repair it or leave it, and shows how the ending depends less on whether the alliance ever dipped than on whether the dip was recovered.

Mechanisms of Change

That the alliance predicts outcome does not establish that it produces it, and disentangling the two is the central problem of process research. Alan Kazdin argued that after decades of demonstrating that therapies work, the field's pressing task is to show how they work, to identify the mediators and mechanisms that carry the effect from treatment to outcome, and he set out the evidential requirements, a demonstrated association, a temporal precedence of the mechanism before the change, and a plausible, specific causal pathway, that most alliance research had yet to meet (Kazdin, 2007). The difficulty is concrete. If clients who improve early then report a better alliance, an alliance-outcome correlation could arise because improvement builds alliance rather than the reverse. Sigal Zilcha-Mano reframed the question by distinguishing two things a single alliance score confounds: a client's stable, trait-like capacity to form good relationships, which they bring to therapy and which predicts outcome without being caused by it, and the state-like, within-person strengthening of the alliance over the course of treatment, which is the part a therapist can actually influence and the part most plausibly therapeutic (Zilcha-Mano, 2017). On this account earlier studies that simply correlated alliance with outcome could not tell the two apart, and the causal signal lies in the within-person change, not the between-person difference. A parallel line of evidence attacks the same problem from the therapist's side. Scott Baldwin and colleagues partitioned the alliance into the variation due to patients and the variation due to therapists and found that it was the therapist component of the alliance that predicted outcome: therapists who formed better alliances across their caseloads had better-improving clients, whereas a patient's own deviation from their therapist's average alliance did not predict their outcome, pointing to the alliance as something a skilled therapist contributes rather than merely a byproduct of an easy client (Baldwin et al., 2007). Understanding change also means understanding the events within sessions, the emotional processing and meaning-making that task-analytic process research was designed to model (Greenberg, 1986). The demonstration below builds Bordin's three-part alliance from its components and shows how the whole is held back by its weakest bond.

Empathy and Expectation

The alliance is the most measured of the relational processes but not the only one that carries weight. Robert Elliott and colleagues, updating the meta-analysis of therapist empathy, found empathy to be a moderate but consistent predictor of outcome, of a magnitude comparable to the alliance and robust across theoretical orientations, vindicating the emphasis Rogers had placed on accurate empathic understanding half a century earlier (Elliott et al., 2018). Alongside what the therapist offers is what the client expects. Michael Constantino and colleagues, meta-analysing studies of outcome expectation, found that clients who entered treatment believing it would help tended to fare better than those who did not, a small but reliable association that makes expectation a genuine therapeutic process rather than a mere placebo nuisance to be controlled away (Constantino et al., 2018). These findings do not stand in isolation; they are part of a broader programme, the interdivisional effort to identify the elements of the therapy relationship that demonstrably work, which concluded that several relational processes, the alliance, empathy, positive regard, collaboration, and the management of expectation, are each supported well enough to be considered evidence-based in their own right, and that tailoring the relationship to the individual client is itself a demonstrably effective practice (Norcross & Lambert, 2018). The upshot is a picture in which the relationship is not the backdrop against which technique operates but a set of active, separately evidenced ingredients.

Worked Example

The alliance-outcome correlation is easy to underrate because r = 0.28 sounds small, and the binomial effect-size display is the standard device for reading it honestly. The display converts a correlation into a difference between two success rates by the simple rule that the success rate in the more favourable group is 0.50 plus half the correlation and the rate in the less favourable group is 0.50 minus half the correlation. Take the alliance-outcome correlation of 0.28 reported in the largest synthesis (Flückiger et al., 2018). Half of 0.28 is 0.14. So among clients with a stronger-than-average alliance the improvement rate is 0.50 plus 0.14, that is 0.64, or 64 percent, while among clients with a weaker-than-average alliance it is 0.50 minus 0.14, that is 0.36, or 36 percent. The correlation that looked negligible corresponds to improvement in nearly two-thirds of one group against barely more than a third of the other, a 28-percentage-point difference in the chance of getting better. The variance-accounted-for framing tells the opposite-seeming story from the same number: 0.28 squared is about 0.078, so the alliance statistically accounts for roughly 8 percent of the variance in outcome. Both figures are correct, and the tension between a mere 8 percent of variance and a 28-percentage-point swing in recovery is the whole lesson: a percentage of variance understates the practical size of a small correlation, whereas the recovery-rate framing makes it legible. Neither figure asserts that the alliance is the cause; the correlation is compatible with the alliance driving improvement, with early improvement building the alliance, or, most likely, with both operating at once (Zilcha-Mano, 2017).

Current Directions

The live frontier of process research is causal, not descriptive. The field now largely agrees that the alliance and outcome are associated; what it is working to establish is which part of that association reflects something a therapist does. The trait-versus-state decomposition has become the organising idea, with newer designs using session-by-session, within-person measurement to isolate the state-like alliance change that a between-person correlation cannot see, and treating a client's stable relational capacity as a variable to be modelled rather than noise to be pooled (Zilcha-Mano, 2017). A second direction turns from the average effect to the therapist who produces it, following the finding that therapist-level differences in alliance predict outcome to ask what more effective therapists actually do, and whether alliance-fostering skill can be trained (Baldwin et al., 2007). A third extends the evidence base for the relationship as a set of separable, individually testable processes and presses the question of how the relationship should be adapted to the particular client, moving from whether the relationship matters to which relational process helps, for whom, and when (Norcross & Lambert, 2018). Running through all three is a methodological shift from the single post-hoc alliance score toward intensive longitudinal measurement, the design best suited to catch a process as it unfolds rather than after it has finished (Flückiger et al., 2018).

Discussion

Psychotherapeutic processes matter first because they reframe what a therapy is. If treatments with sharply different techniques reach broadly similar outcomes, then the causal weight cannot lie mainly in the technique, and attention shifts to the relational and expectational processes the treatments share (Wampold, 2015; Cuijpers et al., 2019). They matter second because they supply psychotherapy with its most durable empirical regularity: the alliance-outcome correlation is modest but has survived decades of replication across disorders and orientations, which is more than can be said for many specific-ingredient claims (Flückiger et al., 2018). They matter third because they pose a hard and general problem in causal inference, the problem of telling a genuine mechanism from a marker of progress already underway, and the responses to it, the trait-state decomposition and the partition of alliance variance into patient and therapist components, are models of how to sharpen a correlational finding toward a causal one (Zilcha-Mano, 2017; Baldwin et al., 2007). And they matter in practice, because the processes that carry change, forming and repairing the alliance, conveying empathy, cultivating realistic hope, are trainable, so that identifying them is not an academic exercise but a route to doing therapy better (Norcross & Lambert, 2018).

Common Misconceptions

The specific technique is what makes a therapy work.

Bona fide therapies built on different techniques tend to produce comparable outcomes, and the common relational processes account for a larger and more reliably measured share of the effect than the ingredients unique to any one brand (Wampold, 2015; Lambert & Barley, 2001). Technique matters, but largely as the vehicle for those shared processes.

A good alliance means the absence of conflict.

Ruptures in the alliance are common and are not in themselves a bad sign; it is the repair of a rupture that is associated with better outcome, so a therapy in which strains arise and are mended can end better than one that never openly strained (Safran et al., 2011; Eubanks et al., 2018).

A strong alliance obviously proves the alliance heals.

The alliance-outcome correlation is compatible with the alliance causing improvement, with early improvement building the alliance, or both; separating a client's stable relational capacity from the therapist-driven change in the alliance is exactly what a bare correlation cannot do (Kazdin, 2007; Zilcha-Mano, 2017).

Glossary

Abreaction.

The discharge of emotion attached to a previously repressed experience as it is brought back into awareness, one of the MeSH subtypes of psychotherapeutic process.

Binomial effect-size display.

A device that converts a correlation into two success rates, 0.50 plus and minus half the correlation, to express a small correlation in terms of a difference in outcome rates.

Bona fide therapy.

A treatment delivered by a trained therapist, based on a coherent psychological rationale, and intended to be therapeutic; the class of therapies whose broadly comparable outcomes motivate the common-factors account.

Change process research.

The study of the moment-to-moment events within sessions through which therapy produces change, as opposed to outcome research asking only whether a treatment works.

Common factors.

The therapeutic ingredients shared across treatments, such as the alliance, empathy, and expectation, held to account for much of the benefit common to all bona fide therapies.

Contextual model.

Wampold's account of how psychotherapy works, through a genuine relationship, the client's expectation of help, and engagement in specific therapeutic actions.

Corrective experience.

A new emotional experience, such as a relationship that survives conflict, that revises a client's prior expectations; the proposed mechanism by which rupture repair helps.

Dodo bird verdict.

The claim, named for the Dodo's pronouncement in Alice in Wonderland, that bona fide psychotherapies produce broadly equivalent outcomes; the equivalence finding the common-factors tradition exists to explain.

Empathy.

The therapist's accurate understanding of the client's experience and perspective; one of Rogers's proposed conditions for change and a moderate, consistent predictor of outcome in its own right.

Mechanism of change.

The process that carries a treatment's effect to its outcome, distinguished from a mere correlate by evidence of temporal precedence and a specific causal pathway.

Outcome expectation.

A client's belief on entering treatment that it will help; a small but reliable predictor of how well the client fares.

Rupture.

A deterioration in the collaborative relationship, a breakdown in the bond or a disagreement on tasks or goals, ranging from a subtle withdrawal to an open confrontation.

Therapeutic alliance.

The collaborative and affective bond between client and therapist together with their agreement on goals and tasks; the most studied psychotherapeutic process.

Trait-state distinction.

The separation of a client's stable capacity to form good relationships from the within-treatment change in the alliance, used to isolate the therapeutically active component.

Transference.

The client's redirection onto the therapist of feelings originally attached to significant figures from the past, a classical psychotherapeutic process.

Working alliance.

Bordin's pantheoretical construct comprising the bond, agreement on goals, and agreement on tasks; a synonym for the therapeutic alliance emphasising its collaborative, cross-theoretical nature.

Key Researchers

Christoph Flückiger. Professor of clinical psychology at the University of Kassel, having previously worked at the Universities of Zurich and Bern; he led the largest meta-analytic synthesis of the alliance-outcome relationship and works on mechanisms of change. Faculty Page - ORCID - Google Scholar - Wikipedia

Leslie S. Greenberg (b. 1945). Distinguished Research Professor Emeritus at York University; the founder of emotion-focused therapy and a pioneer of change-process research, which studies the moment-to-moment events within sessions through which therapy produces change. ORCID - Wikipedia

Carl R. Rogers (1902-1987). Founder of client-centered therapy and, with his 1957 statement of the necessary and sufficient conditions for change, the originator of the relational, common-factors tradition in psychotherapy. Wikipedia

Jeremy D. Safran (1952-2018). Professor of psychology at the New School for Social Research; he originated the research programme on alliance ruptures and their repair, recasting relational strain as a therapeutic opportunity. Wikipedia

Bruce E. Wampold (b. 1948). Emeritus professor at the University of Wisconsin-Madison and, at Modum Bad in Norway, a leading proponent of the contextual model and of the meta-analytic case for the common factors. ORCID - Google Scholar - Wikipedia

Sigal Zilcha-Mano. Full Professor of psychology at the University of Haifa; she leads work on mechanisms of change and the trait-state decomposition of the alliance, sharpening the question of whether and how the alliance is therapeutic. Faculty Page - ORCID - Google Scholar

Frequently Asked Questions

What are psychotherapeutic processes?
Psychotherapeutic processes are the change-producing events that unfold within therapy, the relationship between client and therapist, their shared goals and tasks, moments of strain and repair, and the client's engagement, expectation, and emotional processing, as distinct from the named techniques of any particular school (Greenberg, 1986). They are studied because different techniques often produce similar outcomes, which points to the shared processes as where much of the change is carried (Wampold, 2015).

What is the working alliance?
The working alliance is the collaborative relationship between client and therapist, which Edward Bordin defined as having three parts: the emotional bond, agreement on the goals of treatment, and agreement on the tasks used to reach them (Bordin, 1979). Bordin's contribution was to make the alliance a pantheoretical construct applying to every therapy rather than only the psychoanalytic one in which it arose.

How strongly does the alliance predict outcome?
Across more than three hundred studies the alliance-outcome correlation is about r = 0.28, modest in size but one of the most robust and replicated findings in psychotherapy research, holding across disorders and treatment types (Flückiger et al., 2018). The first meta-analysis of the question reached the same qualitative conclusion decades earlier (Horvath & Symonds, 1991).

Is a correlation of 0.28 large enough to matter?
Read as a binomial effect-size display, a correlation of 0.28 means the improvement rate rises from about 36 percent in clients with a weaker alliance to about 64 percent in those with a stronger one, a 28-percentage-point difference in the chance of getting better (Flückiger et al., 2018). It accounts for only about 8 percent of outcome variance, so the same number can look small or large depending on how it is expressed.

Are the common factors all that matter, or does technique count?
The common-factors position is not that technique is inert but that the shared relational and expectational processes account for a larger, more reliably measured share of the benefit than the ingredients unique to any one brand (Lambert & Barley, 2001). Wampold's contextual model treats technique as the vehicle through which the relationship, expectation, and specific actions operate (Wampold, 2015).

What is an alliance rupture, and are ruptures bad?
A rupture is a deterioration in the collaborative relationship, a breakdown in the bond or a disagreement on tasks or goals, and it is common rather than a sign of failure (Safran et al., 2011). Repaired ruptures are associated with better outcomes, so a therapy in which strains arise and are mended can end better than one in which they are avoided or ignored (Eubanks et al., 2018).

Does the alliance actually cause improvement?
This is unsettled. A correlation between alliance and outcome is compatible with the alliance driving improvement, with early improvement building the alliance, or both, and establishing genuine causation requires showing the mechanism precedes the change through a specific pathway (Kazdin, 2007). Separating a client's stable relational capacity from the therapist-driven change in the alliance is the current strategy for isolating the causal part (Zilcha-Mano, 2017).

Besides the alliance, which relational processes are supported by evidence?
Therapist empathy is a moderate and consistent predictor of outcome, comparable in size to the alliance (Elliott et al., 2018), and a client's expectation that treatment will help is a small but reliable one (Constantino et al., 2018). A broad review concluded that several relational elements, the alliance, empathy, positive regard, collaboration, and expectation, are each supported well enough to count as evidence-based (Norcross & Lambert, 2018).

References