Contribution
The paper does three things. It gives a short history of behaviour therapy and CBT as a cycle of empirically grounded treatment development (phenomenology → theory of maintaining factors → experiment → treatment → trial; Clark’s six steps). It argues that the field has confused common factors with mechanisms of change. And it proposes a single mechanism of change for effective psychological therapy in general. The proposal is the new part. The rest is an editor’s synthesis for an anniversary issue.
Key insight
Psychological problems are, in the paper’s view, “exaggerations of one or more otherwise normal processes”, sustained not by their original causes but by maintaining factors. The central one is a meaning: the panicking person reads palpitations as a heart attack; the OCD father reads an intrusive thought as proof he is dangerous. That meaning drives behaviour (avoidance, neutralising, safety-seeking), and the behaviour keeps the meaning from being disconfirmed. Therapy works when it gives the person a credible alternative account and a way to test it. The paper’s slogan is “don’t trust me, find out for yourself”. If that is the mechanism, then exposure, behavioural activation and even insight therapies work to the extent that they deliver a tested Theory B, explicitly or not.
Assumptions
- Maintenance, not cause. “The past 50 years have seen no significant development of our understanding of causal factors in mental health problems” (p. 600). Treatment development therefore targets maintaining factors (p. 600; Clark’s step 2, p. 601).
- Continuity with normal processes. Social anxiety continues shyness; depression continues ordinary low mood (p. 600).
- Appraisal drives emotion. People “experience mood problems because they interpret their situation as being more negative or threat-based than it really is” (p. 605). This is stated as the conclusion of “decades of psychological research”, without specific citation at that point.
- Two criteria for a credible alternative. It must fit the person’s past experience and survive their future experience, incidental or experimental (p. 605).
Key results
No new data. The results are arguments and a proposal, with cited evidence.
- The logic of treatment evidence (p. 599). “The effectiveness of a treatment is not evidence for the validity of the theory on which it is based with one important exception”: when a treatment demonstrably changes a factor crucial to a theory but does not change the target problem, that disconfirms the theory. Successful, theory-informed treatment is only “consistent with” its theory.
- Common factors versus mechanisms (p. 603). “Although mechanism of change should necessarily manifest in the occurrence of common factors in therapy, common factors do not necessarily reflect mechanisms of change.” The alliance may follow improvement rather than drive it.
- The exposure circle (p. 598). Exposure was adopted as the mechanism of anxiety treatment because it was common to successful treatments, and inferred from fear reduction. The paper calls this circular, and says the confusion of common factors with mechanisms “has in our view held the field back”.
- Why mechanisms remain unclear (p. 602). The paper gives four reasons: the emphasis on efficacy studies; a lack of statistical procedures linking independent and dependent variables; a near-total absence of studies establishing the timeline of change; and failure to test CBT-specific interventions against general ones.
- The case for cognitive change (pp. 603–604). Against the objections from component analyses and from early sudden gains, the paper relays Lorenzo-Luaces et al. (2015): only three studies met their four questions with temporality controlled, and those authors argue cognitive change is the mechanism, whatever procedure produces it.
- The proposal (pp. 605–608). The core mechanism is “the explicit identification of the problematic negative meaning (‘Theory A’) and its contrast to the shared understanding collaboratively developed as part of formulation (‘Theory B’)” (p. 607). The alliance is “necessary insofar as it facilitates the development of a shared understanding … but it is not sufficient to engender the necessary psychological flexibility in terms of meaning making” (p. 608). Effective non-cognitive therapies are said to “implicitly develop alternatives” (p. 606), and all psychological therapies are “underpinned by a more or less explicit Theory B” (p. 607).
- Cited support for the proposal (pp. 606–607). Exposure presented as behavioural experiments designed to disconfirm threat beliefs “was very markedly superior to exposure alone” in agoraphobia (Salkovskis et al. 1991, 1999) and social phobia (Wells et al. 1995). Panic attacks undid habituation (Matthews et al. 1981).
- Agenda (pp. 608–609). Measure mechanisms alongside efficacy; use mediation statistics; compare specific with general interventions; and establish temporal precedence by Kazdin’s (2007) six criteria. Those are that the treatment changes the mediator, which then changes the outcome; that the mediator–outcome relation is specific; that other mediators are ruled out experimentally; consistency; a gradient; and multiple data sources. Ecological momentary assessment and sudden-gains methods are recommended.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | A treatment’s effectiveness is not evidence for its theory, except that changing the theory’s crucial factor without improvement disconfirms it | strong as a logical point | p. 599 |
| C2 | Common factors need not be mechanisms; the alliance may be a consequence of improvement | moderate: argued, with the shoes example; the reverse direction is suggested, not shown | p. 603 |
| C3 | No significant progress has been made on causes of mental health problems in 50 years; treatment has advanced by targeting maintaining factors | moderate: a judgement of the field, with risk-factor work noted | p. 600 |
| C4 | Neuroimaging has failed to generate useful findings for understanding mental health problems | asserted with four citations; strong wording, not argued here | p. 600 |
| C5 | Exposure framed as belief-disconfirming behavioural experiments is markedly superior to exposure alone | moderate: three cited studies, all co-authored by Salkovskis | p. 607 |
| C6 | The mechanism of change in all effective psychological therapies is a shift from a stuck negative meaning to a tested alternative, a type of psychological flexibility | weak-to-moderate: a proposal argued from CBT’s history and clinical cases; not tested against rival mechanisms | pp. 605–608 |
| C7 | The alliance is necessary but not sufficient for that shift | interpretive | p. 608 |
| C8 | Mechanism research has largely failed Kazdin’s criteria, especially temporal precedence | moderate: cited reviews | pp. 608–609 |
Concepts
- mechanism of change (MOC): what actually produces therapeutic change. It is distinguished from a common factor, which is merely present across therapies.
- common factors / common elements: features shared by therapies, whether relational (the alliance) or topographical (similar techniques).
- Theory A / Theory B: the person’s threat-laden account of their problem, and the collaboratively formulated, testable, less threatening alternative.
- psychological flexibility (as used here): being able “to understand and weigh different perspectives on what they are experiencing” (p. 605), and then to respond more flexibly. It is explicitly not thinking “more rationally or logically”. The paper cites none of the ACT literature that uses the same phrase.
- descriptive vs mechanistic transdiagnostic processes (Harvey et al. 2011): seen in many disorders without a shared functional role (low self-esteem), versus maintaining at least two disorders (preoccupation with shape and weight).
- empirically grounded clinical interventions: the paper’s term for CBT as a cycle of clinical phenomenology, theory, experiment and outcome research (Figure 1).
Connections
- Process-based therapy (LIT-544). Both papers open with Gordon Paul’s question. This one uses the 1967 form (”… and under which set of circumstances?”). Hofmann and Hayes use the 1969 form, which adds “and how does it come about?”, and say Paul himself added that clause. Both diagnose an efficacy-first field that neglected mechanism, and both cite Kazdin (2007) and Nock (2007). They differ on the answer. Hofmann and Hayes expect many processes, arranged hierarchically. This paper proposes one mechanism across therapies. The word “flexibility” is common to both, but the referents differ. Here it is meaning-making flexibility, reached by testing beliefs. In ACT, the tradition Hayes leads, “psychological flexibility” is a process that the paper under note does not define or cite. Whether they are one construct is open.
- Network theory (LIT-547). “Vicious circles which trap them” (p. 608) and being “stuck” are the feedback-maintained disorder state of Borsboom’s theory. The difference is the level of the nodes: appraisals and safety behaviours here, manual symptoms there. On Borsboom’s taxonomy, CBT as described here is a network intervention. It weakens the link by which a meaning drives the behaviour that protects it.
- Emotion regulation (LIT-554). The paper asks why mechanism research in CBT has not done what “has been done and replicated in emotion regulation studies” under strict experimental conditions, and cites Gross (1998), Gross and John (2003) and Webb et al. (2012) for it (p. 609). Of these, Gross (1998) is the conceptual review the record holds, which, by its filing note, has no data of its own. The experimental work the paper means must be in the other two. Gross’s “cognitive change” family of regulation strategies (reappraisal) is close kin to the shift from Theory A to Theory B. The paper does not draw that link.
- Paul’s question as a shared root. Both this paper and LIT-544 treat Paul (1967/1969) as the founding statement of the field’s goal.
Bearing on the record
- Proposed THEORY material. C6 is the paper’s thesis, but one source stating its own proposal cannot carry an Active theory. The batch report proposes a broader convergence that this paper supports: disorder is maintained by self-reinforcing loops, and treatment works by breaking them.
- No ML instruction. Nothing here belongs in the anthology.
- C1 is filed, with LIT-544, as THEORY-053 (Active). The maintenance loop is one source of THEORY-058.
Limitations
- The mechanism is not tested here, and by the paper’s own agenda (Kazdin’s criteria) the evidence that would test it is mostly not yet collected.
- “All effective therapies” is argued by reinterpretation. That exposure, behavioural activation or insight therapy work through an implicit Theory B is asserted. A rival account (habituation, inhibitory learning, reinforcement) is not shown to fail.
- The appraisal premise is stated as the lesson of “decades of research” without specific citation at that point.
- Interest. The first author is the journal’s editor, the paper is part of an anniversary issue he co-introduced, and all three studies cited for the key comparison (C5) have him as an author (Salkovskis et al. 1991, 1999; Wells, Clark, Salkovskis et al. 1995). The paper discloses the editorship and states he was not involved in its review.
- Unanchored terminology. The paper takes “psychological flexibility” without engaging the measures and literature that already use the term.
Open questions
- Does a measured shift in belief in Theory A versus Theory B precede, and predict, symptom change across therapies, including non-CBT ones?
- Is the paper’s meaning-making flexibility the same construct as ACT’s psychological flexibility, or a narrower one?
- If disorders are network states (LIT-547), is a change of meaning one network intervention among several, or the one that generalises?
Corrections
- none to a seeded skim (there was no seed)
- Citation year. The brief’s “2024” is the first-online year (5 January 2024). The issue is dated November 2023, and the article’s own citation line gives 2023.