How the Major Modalities Recalibrate Priors
A predictive-processing crosswalk — technique, lever, floor, and when it fails — across CBT, ACT, DBT, psychodynamic/EFT, EMDR, and MI
Companion to A Unified Clinical Model of Psychotherapy. The paper argues that the hundred-odd therapies work by reaching the same machinery through different doors. This note walks through the doors. For each major modality it names the core logic in predictive-processing terms, lays the signature techniques out as a table of technique → lever pulled → floor reached → when it fails, and closes with the one thing the frame adds that the modality’s own manual doesn’t. Read it as a map for cross-training and for troubleshooting a stall — not as a claim that any tradition reduces to this vocabulary. Held by the therapist; matched to the client’s floor via the Decision Aid — Locating the Floor. The six modalities below get a full section each; the complete counseling-theories canon — Corey’s eleven — is mapped in a single table near the end.
Every technique below pulls one or more of the same small set of levers the engine runs on:
Content — revise what the prior claims (change the belief itself).
Precision — re-weight confidence: turn the stuck prior down, the disconfirming evidence up. The valve.
Prediction error — engineer one real, unmanufactured mismatch between prediction and outcome.
Furnace — interrupt the behavior quietly manufacturing the confirming evidence (provoke · avoid · filter · reinterpret · withdraw effort).
Reconsolidation — reactivate a consolidated emotional prior so it goes briefly labile, then update it in the presence of a new outcome (the floor 6–7 special case).
Valve / relationship — the safety, belonging, and standing that let any error count at all.
The when it fails column is the point of the whole exercise. A crosswalk that only renamed techniques in new words would be a thesaurus. The frame earns its keep only where it predicts why a sound technique bounces off a particular client — and that is almost always a lever–floor mismatch, an uninterrupted furnace, or a shut valve. If a row’s failure mode is just “when you do it badly,” it isn’t pulling its weight.
1 narrative · 2 metacognition · 3 beliefs · 4 relational template · 5 scripts/sequences · 6 procedure–emotion · 7 body · 8 cultural substrate. Glass (verbal, 1–3) vs concrete (embodied, 5–7). Full map in A Unified Clinical Model of Psychotherapy Part 3 and the Decision Map — Locating the Floor.
The core six, side by side
| Modality | Primary lever(s) | Home floors | Its distinctive reach |
|---|---|---|---|
| CBT | Content + prediction error | 3, 5 | Falsifying a false belief with engineered evidence |
| ACT | Precision + furnace | 2, 6–7 | Recalibrating a true prior without touching its content |
| DBT | Precision management + valve | 7, conditions | Cooling a hot precision system enough to work at all |
| Psychodynamic / EFT | Relational prediction error + reconsolidation | 4, 6 | The live corrective emotional experience |
| EMDR | Reconsolidation | 6 | Destabilizing a maximally consolidated memory prior |
| MI | Precision (source of evidence) + valve | change process | Moving ambivalence without firing the furnace |
Two things fall out of the table. First, no modality owns all the levers, and none is complete — which is the map’s practical payoff. Pick the tradition whose lever fits the floor the client’s problem sits on; when your current approach stalls, read across the when it fails columns to find the lever you’re not pulling. A CBT that won’t move is often an EFT-shaped problem — a floor-6 prior that needs reactivation, not evidence. An MI that stalls may be a genuine floor-7 dysregulation that needs DBT’s precision-down before any change-talk can be weighted at all.
Second, the convergence runs deepest exactly where the theories fight hardest. The easy agreements — exposure is prediction error — a skeptic can wave off as relabeling. The hard one — ACT, whose philosophy of science forbids the representational vocabulary the frame is built on, converging anyway on change the relationship to the thought, not its content — is the one that can’t be dismissed. That is the argument the crosswalk exists to make.
The six above are the modalities you actively run — each deep enough to execute straight from the table. But the lens is meant to cover the whole field, so here is the other cut: every named approach in the standard counseling-theories curriculum, mapped one row each.
The counseling-theories canon (Corey, ch. 4–14)
These are the eleven approaches a counseling student meets in the survey course; the chapter numbers track Corey’s Theory and Practice of Counseling and Psychotherapy. The paper’s claim is that one lens accounts for all of them — not by flattening them, but by showing which lever each one pulls and which floor it reaches. Read one thing first: two of these barely fit the phrase “recalibrate a prior,” and that is the point, not a gap — some approaches refuse to treat a calibrated prior at all. The framework is honest about where the work is not error-correction.
| Modality (ch.) | What it is, in PP terms | Primary lever(s) | Home floor(s) | Where it stalls / what to watch |
|---|---|---|---|---|
| Psychoanalytic (4) | Transference and interpretation surface a defended floor-4 relational template; resistance is the furnace protecting it | Relational prediction error; making priors legible | 4, 6 | Intellectual insight without the live relational error is inert (run deep in the Psychodynamic & EFT section above) |
| Adlerian (5) | The “style of life” is a narrative goal-prior (private logic, fictional finalism); “acting as if” is a behavioral experiment; “spitting in the soup” names the hidden payoff so the furnace move stops working | Narrative content + prediction error + furnace-interrupt | 1, stakes (belonging) | Encouragement with no real test stays inspirational; the belonging need (sociometer) must be named, not assumed (full section below) |
| Existential (6) | The “givens” — death, freedom, isolation, meaning — are calibrated priors with nothing to falsify; you work the stance and the meaning-frame | Narrative / meaning + valve (authentic presence) | 1, conditions | Treating a true existential prior as a distortion to fix — the calibration gate at the scale of a life |
| Person-Centered (7) | Rogers’ core conditions (empathy, UPR, congruence) are the valve; and when the prior is relational, the regard that survives contact is the disconfirming evidence | Valve + relational prediction error | conditions; 4 (when the regard is put at risk) | When it is only comfortable — a regard never at risk teaches a floor-4 prior nothing, and the therapy stays on the glass floors |
| Gestalt (8) | “Unfinished business” is an un-updated floor-6 emotion held out of awareness; the empty chair reactivates it live | Reconsolidation + furnace-interrupt + precision on the body | 6, 7, 2 | Enactment without safety re-runs the wound; the concrete-floor cousin of EFT (full section below) |
| Behavior (9) | Conditioned fears and contingencies are floor-5/6 action-priors; exposure is the pure prediction-error engine | Prediction error + furnace-interrupt; contingency = script revision | 5, 6 | Safety behaviors leave the furnace running — it’s the error engine inside CBT, without the content layer |
| Cognitive Behavior (10) | Revises the content of a floor-3 belief and engineers a real error to test it | Content + prediction error | 3, 5 | Words aimed at a concrete prior (run deep in the CBT section above) |
| Choice Theory / Reality (11) | Active inference in plain language — you act to make the world match your “quality world” of wanted-priors; WDEP’s self-evaluation is a self-generated discrepancy | Self-generated prediction error + new script (planning) | 1 (wants), 5 (doing) | Self-evaluation that shames instead of revealing the gap; it needs the valve MI supplies (full section below) |
| Feminist (12) | Symptoms are often calibrated reads of real power and oppression; the pathogen is frequently the floor-8 substrate, not the individual prior | Name the substrate; calibration gate; egalitarian valve | 8, conditions | Privatizing a structural injury as a personal distortion — the institutional reason the calibration gate exists |
| Postmodern — SFBT & Narrative (13) | No single objective prior to correct; hunt the disconfirming evidence the problem-story filters out (exceptions, unique outcomes) and re-author; externalizing is defusion at the level of identity | Narrative content + precision (attention to exceptions) + defusion | 1, 2 | A future-prior (the miracle question) with no path built to it; re-authoring over an unregulated concrete floor (full section below) |
| Family Systems (14) | The furnace scaled to a whole network — each member manufactures the others’ predicted response, so the loop maintains the symptom, not the person | Systemic furnace-interrupt + new error via enactment; reframing = precision / narrative | 4 (at system scale), 5 | Working the identified patient alone can’t reach a system-maintained prior; you have to change the loop (full section below) |
The canon sorts into four ways in — plus the three already run deep above.
The calibration gate — Existential, Feminist.
These two barely fit “recalibrate a false prior,” which is exactly why they matter. Existential insists many priors are true — you don’t falsify death, you re-author the meaning around it. Feminist insists the pathogen is often the floor-8 substrate, not the person — the institutional form of check the field before treating the prior. Together they are the framework’s conscience: not every prior is a distortion, and the world does much of the work.
The valve that is also the evidence — Person-Centered.
Rogers’ conditions are the common-factors spine the whole paper leans on: the safety that lets errors already arriving finally count. But the account stops short if it stops there. When the stuck prior is itself relational — what closeness costs, what happens when you need something — the therapist’s regard is not only the channel. A person over time is the only evidence that bears on such a prior, and regard that was free to drop and did not is the error itself. One object fills both roles, which is what Rogers was reporting when he called the conditions necessary and sufficient: sufficient for the priors his caseload was full of, and not for a contamination prior or a bodily one. That also relocates where it fails. Not “no aimed error” — only comfortable. A regard never at risk teaches a floor-4 prior nothing.
Active inference, said in plain language — Adlerian, Reality Therapy.
Both picture a person acting to make the world match a wanted-model, and both locate change in self-evaluating the gap. Adler’s “style of life” is a generative model and his “social interest” is the sociometer; Glasser’s “quality world” is a bundle of goal-priors and his self-evaluation question is MI’s develop-discrepancy without the Rogerian wrapper. They described the engine decades before the vocabulary existed.
Reaching the concrete floors by enactment — Gestalt, Family Systems.
Neither works by talking about the problem. Gestalt’s empty chair reactivates a floor-6 emotion live — the direct ancestor of EFT’s reconsolidation move — and Family Systems is the furnace scaled to a network, where the confirming evidence is manufactured by other people, which is precisely why individual insight can’t reach it and you have to change the loop.
Working the story — Postmodern (SFBT & Narrative).
These formalize the fight against one specific furnace move, the filter. A problem-saturated story throws away every exception; SFBT and Narrative systematically hunt that discarded disconfirming evidence — exceptions, unique outcomes — and re-weight it. Narrative’s externalizing (“the problem is the problem; the person is not the problem”) is ACT defusion performed at the level of identity.
Already run deep above — Psychoanalytic, Behavior, CBT.
Corey’s Chapter 4 is the Psychodynamic & EFT section; Chapter 9 Behavior is the pure prediction-error engine sitting inside CBT (exposure without the content layer); Chapter 10 CBT is that same CBT section. They pin the historical canon to the six workhorses — same machinery, older names.
Beyond both lists, the modern process modalities — IFS, MBCT, somatic and sensorimotor work — map the same way through the floor→family column of the Decision Aid — Locating the Floor.
The canon approaches, up close
Five of the eleven earn the full treatment the core six get, because they are genuinely technique-rich and their techniques map cleanly onto the levers. Gestalt and Family Systems reach floors talk can’t, by enactment. Adlerian, the two Postmodern approaches, and Reality Therapy each formalize one lever — a prediction-error experiment, a reversal of the filter, a self-generated discrepancy. (The remaining entries are either already deep above — Psychoanalytic, Behavior, CBT — or, like Person-Centered and Existential, deliberately not technique-driven, where a technique table would misrepresent the method — for those two the relationship and the stance are the method, which is precisely why the table fails them; both are mapped in the clusters above.)
CBT — the clean case
Core logic in PP. CBT is the modality predictive processing describes most transparently, because it does openly what the frame says every therapy does covertly: it revises the content of a prior (floor 3) and engineers a real prediction error (floors 3 and 5). Restructuring goes at the claim; behavioral experiments and exposure go at the evidence.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Thought record / cognitive restructuring | Content + precision — surface the claim, widen the evidence base, loosen confidence | 3 | The belief is held below the words — “I know it’s not true and I still feel it.” Accurate insight that changes nothing means a floor-3 tool aimed at a floor-6/7 prior. Also fails when the prior is calibrated — the evidence really does support the fear. |
| Behavioral experiment | Prediction error — turn the belief into a falsifiable prediction, run it, compare | 3 / 5 | The furnace manufactures the outcome: the client runs the test but provokes, filters, or withdraws effort so it confirms the fear. Or the worst case isn’t survivable, so they (rightly) won’t run it. |
| Graded exposure | Prediction error + furnace-interrupt (blocks avoidance) | 5 | Safety behaviors left in place. The disconfirmation gets credited to the safety behavior — “I only got through it because I gripped the rail” — so the prior never takes the hit. |
| Behavioral activation | Prediction error against “nothing helps / I can’t feel good” + feeds the mattering input stream | 3 / 5 | In status-injury depression it can be metabolized as more failed striving — one more climb the world doesn’t reward. Privatizes a structural loss; pair with the mattering moves in Clinical Applications — Status, Mattering, and Rank. |
What PP adds. A diagnosis for the bounce. When a clean, correct thought record changes nothing, CBT’s own manual tends to say do it again, better; the frame says you have the wrong floor — stop feeding evidence to a level that isn’t listening and change channels. It also reframes exposure’s mechanism from habituation to expectancy-violation (a move CBT’s own science made independently), which tells you the load-bearing steps are the written prediction and the blocked safety behavior — not the dwell time.
CBT-I — the worked case: one protocol, every lever
Core logic in PP. Chronic insomnia is the frame’s cleanest everyday demonstration, because the perpetuating machinery is almost never the beliefs. Conditioned arousal — the bed itself now a cue for wakefulness — is a floor-5/6 prior firing before thought. The catastrophic sleep beliefs run as forecasts that generate the arousal they predict (floor 3). Hypervigilant monitoring — clock-watching, body-scanning for sleep — is precision misallocated, amplifying exactly what it checks. And sleep effort is the master furnace move: naps, early bedtimes, hours awake in bed, all manufacturing the confirming evidence nightly. CBT-I unbundles into the levers almost without remainder — and sleep cannot be placed under full voluntary control, so that effort under load becomes counterproductive, which the frame reads as the same law the mattering gauge runs on: effort is the interference, not the instrument.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Sleep restriction | Prediction error, engineered — compress time in bed until homeostatic pressure forces fast onset; "I cannot sleep" meets an undeniable miss the client’s own body produced | 5–6 | The first-week dip goes unnamed: worse-before-better reads as proof of failure and the client bails exactly as the mechanism engages. Predict the dip in writing — the extinction-burst rule. Also fails against unscreened biology (apnea, restless legs, circadian disorder): wrong organ. |
| Stimulus control | Furnace-interrupt + path-shaping — out of bed when awake, bed only for sleep; blocks the nightly manufacture of bed = wakefulness | 5–6 | Half-run: the client stays in bed “resting,” and pays the old association on an intermittent schedule — the most durable schedule there is. |
| Cognitive restructuring of sleep beliefs | Content + precision on the catastrophic forecast (“tomorrow is ruined”) | 3 | Accurate insight that changes nothing: the arousal is conditioned, not believed into being, and the bed keeps teaching the body while the words work the wrong floor. |
| Paradoxical intention | Precision — deflate the goal-prior itself; drop sleep effort, since trying produces the arousal that prevents the state | 2 / the goal | Run as a trick: covert effort wearing surrender’s costume — the client tries to not-try and monitors whether it’s working, which is the monitoring again. |
| Relaxation / mindfulness-based variants | Precision down on the monitoring channel — body-scanning and clock-watching | 2 / 7 | Becomes another performance with a score. |
What PP adds. A structure over the package’s own ingredients — the conditioned association and the furnace carry the disorder, at floors five and six, while the component literature refuses to rank the ingredients the way intuition expects: what reliably fails alone is sleep hygiene, not belief work, and the two largest component analyses disagree outright on the belief work’s weight. A name for its modern failure mode: orthosomnia, the tracked and optimized sleeper — the cure become the wall; if the client turns hygiene into a system, that is not diligence, it is the finding. And the gates, applied before any of it: screen the biology first, and check calibration — the newborn, the shift work, the unsafe housing are an environment being accurately read, not a prior to loosen. Fix the field, or name the season, before running experiments at a body that is answering its world correctly.
ACT — the hard case
Core logic in PP. ACT is the sharp case because it does not revise prior content at all. It works two other levers. Defusion drops the precision on a thought-as-literal-fact (floor 2). Acceptance interrupts the furnace’s avoid-move — the struggle against an internal experience — and in doing so updates a second-order prior, I can’t survive this feeling. Values and committed action install a high-level prior (a chosen direction) that gives the predictions below it something to reorganize around. It is the whole engine run without ever touching the belief’s content — which is exactly why it reaches priors CBT can’t.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Cognitive defusion (“I’m having the thought that…”, naming the mind) | Precision — strip the weight off the thought as literal truth | 2 | Used on a thought whose content is false and the situation fixable — defusing from a true, actionable signal becomes spiritualized avoidance. The dividing line: defusion for the calibrated prior, restructuring for the false one. |
| Acceptance / willingness | Furnace-interrupt (ends experiential avoidance) + updates the meta-prior “I can’t survive this feeling” | 6–7 + meta | Misapplied to the situation instead of the internal experience — a client “accepting” a harmful environment that should be changed. Acceptance is of the feeling, never of the abuse. |
| Values clarification | Installs a high-level prior (a chosen direction) that reorganizes what sits below it | 1 / top-down | Floated over an unregulated body or acute risk — you cannot values-clarify through a flashback. Regulate floor 7 first. |
| Committed action | Prediction error — act while predicting you can’t function with the feeling present; the prediction gets violated | 5 | Same boundary as exposure: if the feeling is a true signal of real threat, overriding it is miscalibration, not courage. |
ACT does not rest on neuroscience; it rests on functional contextualism, a philosophy of science that asks of a thought not “is it true?” but “what is it doing?” — and treats the first question as the trap. On that view a thought is a behavior to be understood by its context and history, never an inner representation to be graded for accuracy. So ACT’s own philosophy rejects the representational framing predictive processing assumes: it will not grant that a prior is a claim about the world carrying a precision weight, and it holds that chasing a thought’s content is precisely what keeps clients stuck. The remarkable thing is that the two still prescribe the identical move in the room — change your relationship to the thought, not its truth value (PP: precision-down; ACT: defusion). When even a framework built to refuse your metaphysics performs your moves, that isn’t relabeling; it’s two independent excavations striking the same structure — the strongest corroboration in this note, precisely because the theoretical agreement is absent. (Honest complication for a sharp room: this describes orthodox PP. An enactivist wing — Hutto, Bruineberg, Kiverstein — reads active inference without robust representations, and on that reading the gap to functional contextualism nearly closes.)
What PP adds. For the clinician it explains why defusion works with no argument about the thought’s truth — it moves precision, not content — and why acceptance is curative rather than merely soothing: it interrupts the avoid-move and updates the meta-prior that the feeling is unsurvivable. That yields a placement rule ACT states as philosophy and PP states as mechanics: ACT is the tool for the true or calibrated prior; CBT for the false one. The calibration gate in A Unified Clinical Model of Psychotherapy Part 2.2 is where you decide which you’re facing.
DBT — precision management for a hot valve
Core logic in PP. Emotion dysregulation is, in these terms, a system whose precision estimation itself runs hot — affective and threat signals arrive over-weighted and swamp everything else. DBT is built to get precision back under control long enough for anything else to land. Distress tolerance turns runaway precision down; the skills install new scripts (floor 5); and the central dialectic — radical acceptance and change — is the valve and the error held in the same hand.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Distress tolerance (TIP skills, crisis-survival, self-soothe, distraction) | Precision — rapidly down-regulate an over-weighted affective signal so it stops dominating | 7 | Turned into a life strategy rather than a crisis tool, it becomes the furnace’s avoid-move — the client never stays with a state long enough for a prior to update. DBT knows this: distress tolerance buys time, it doesn’t treat. |
| Check the facts | Content + the calibration gate — is the emotion justified by the facts? | 3 | (The failure-catcher, not a failure site.) DBT’s own version of the calibration check — the discipline the whole model puts first. |
| Opposite action | Prediction error + furnace-interrupt — act opposite to an unjustified emotion’s urge | 3 / 5 | Applied to a justified emotion. DBT builds the guard in: check the facts first; opposite action is only for emotions that don’t fit them. |
| Interpersonal effectiveness (DEAR MAN, etc.) | New script + tests the relational prior “if I ask, I’ll be rejected” | 5 / 4 | The real driver is the floor-4 template, not the missing skill. A script laid over an unexamined attachment prior gets deployed, the furnace re-provokes the feared response, and the skill “doesn’t work.” |
| Dialectics (acceptance + change) | Valve + error together — validation opens the valve, skills deliver the change | conditions | Either half alone. Validation without change is comfort that doesn’t move; change without validation slams the valve shut. The dialectic is the rule PP states as “no error lands through a closed valve.” |
What PP adds. It says why the two wings of DBT are not a stylistic balance but a mechanical necessity: validation lowers precision on threat so the room is safe enough for a signal to count, and only then can a skill deliver the error or install the script. And it locates distress tolerance precisely — precision management, not cure — which predicts its exact failure mode when overused.
Psychodynamic & EFT — the relational and emotional floors
Core logic in PP. These traditions work the floors words can’t reach on their own: the relational template (floor 4, the internal working model) and the consolidated emotional prior (floor 6). Their shared active ingredient is the corrective emotional experience — a real, unmanufactured relational prediction error delivered live inside the alliance — and, in EFT, reconsolidation: making an old emotional prior briefly labile and updating it in the presence of a new attachment response.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Transference / here-and-now work | Prediction error at floor 4 — the template predicts the therapist will judge, control, or leave; the therapist doesn’t, and the error lands live | 4 | Intellectualized. Insight about the transference is floor-1 words about a floor-4 prior — accurate and inert. The mutative version is emotional, not intellectual, insight. |
| Genetic interpretation (linking past to present) | Content / narrative — reorganize the origin story | 1–3 | The textbook “accurate insight that changes nothing” when delivered as explanation without the live relational error. PP’s clearest correction to classical technique. |
| EFT: evoking and deepening primary emotion | Reconsolidation — reactivate the floor-6 prior (the grief or fear under the reactive anger) so it goes labile | 6 / 4 | Evoked without a genuinely new response present — it just re-runs the old emotion and can re-traumatize. The disconfirming outcome has to be there at the moment of lability. |
| EFT: restructuring the negative cycle (couples) | Furnace-interrupt at the dyadic level + new prediction error — the pursue–withdraw cycle is each partner manufacturing the other’s feared response | 4 / 5 | One partner’s prior too high-precision to register the new bid; or safety not yet built (valve shut). De-escalate before restructuring. |
What PP adds. It explains the field’s oldest hard-won lesson in one sentence: emotional insight heals and intellectual insight doesn’t, because intellectual insight edits floor-1 words while the prior sits at floor 4 or 6 — and a floor-4/6 prior updates only when a real, unmanufactured relational error reaches its own level, which requires the emotion to be live at the moment the new outcome arrives. That is the reconsolidation timing rule EFT found empirically and PP states mechanically.
EMDR — reopening a consolidated prior
Core logic in PP. EMDR targets a floor-6 prior held at maximum precision — the traumatic memory network — and appears to work by destabilizing it (reactivation plus a dual-attention task that taxes working memory) so it re-stores with less charge: a reconsolidation account.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Reactivate target + dual-attention (bilateral) stimulation | Reconsolidation — reactivation makes the floor-6 prior labile; the dual task degrades its vividness and affective charge so it re-stores at lower precision | 6 | Run outside a stable-enough window — without dual awareness it’s reliving, not updating, and can re-traumatize. Dissociation and ongoing (calibrated) danger are contraindications. |
| Installation of the positive cognition | Content — pair the desensitized memory with an adaptive belief | 6 → 3 | Attempted before desensitization. Installing a positive cognition over a still-high-precision trauma prior is the “telling them they’re safe” failure — words weighted at zero. Sequence is non-negotiable: precision down first, content second. |
EMDR’s outcomes are well supported; its mechanism is genuinely contested — working-memory taxation, a REM-like reconsolidation process, and plain exposure are all live candidates, and several dismantling studies find the eye-movement component adds little over exposure alone. Predictive processing offers the reconsolidation reading as a candidate that places EMDR in the same family as EFT; it does not settle whether the bilateral stimulation is doing the lifting.
What PP adds. It puts EMDR and EFT in one family — reactivate, destabilize, update — and explains why exposure sometimes needs a destabilizer to move a maximally consolidated prior. And it names the boundary the mechanism implies: dual awareness required, ongoing or calibrated danger a stop.
Motivational Interviewing — precision at the level of change itself
Core logic in PP. MI is the odd one out: its target isn’t a specific prior but the process of changing one at all. Ambivalence is two priors — status quo and change — held at similar precision. MI shifts the weight toward change without supplying any content, and it does so while protecting the valve, because the moment the therapist supplies the argument, the client’s furnace provokes the counter-argument and they talk themselves out of it (the “righting reflex”).
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Evoking change talk | Precision — raise the weight on the client’s own change-prior by having them voice it; self-generated evidence outweighs the therapist’s identical point | 1–3 | No real ambivalence — pure precontemplation, or a calibrated reason not to change. MI can’t manufacture a value that isn’t there. |
| Rolling with resistance / dropping the righting reflex | Furnace-interrupt + valve-protection — don’t hand the client the argument their furnace can push against | conditions | When real information or harm-reduction is genuinely needed and pure reflection withholds it. MI is a stance, not a gag order. |
| Developing discrepancy | Prediction error at the narrative floor — the client’s own “I’m a good father” meets “and I’m doing this,” a self-generated mismatch | 1 | When it lands as shame rather than motivation: the valve slams shut and the furnace defends. Amplify the client’s values, don’t prosecute them. |
What PP adds. MI is the cleanest demonstration that who supplies the evidence is itself a precision variable — the client’s own change-talk is weighted higher than the therapist’s word-for-word identical argument — and that the righting reflex fails for a mechanical reason: argue for X and you fire the furnace that provokes not-X. It shows the valve operating over the whole arc of change, not just a single belief.
Gestalt — error delivered to the body
Core logic in PP. Gestalt reaches the concrete floors (6–7) and present-moment awareness (2) that talk can’t, by enacting the frozen material instead of discussing it. “Unfinished business” is an emotional prior (floor 6) whose update was interrupted and never completed — a prediction error frozen mid-flight. The contact-boundary disturbances (introjection, projection, retroflection, deflection, confluence) are the furnace’s avoid-move working in real time at the boundary between self and world. The method brings the frozen material live into the room, makes it labile, and lets a new completion land as an unmanufactured error — in the body, now.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Empty-chair (unfinished business with an absent figure) | Reconsolidation — reactivate the frozen floor-6 emotion live and let a new completion land as real error | 6 / 4 | Run as role-play performance without genuine affective arousal — the prior never goes labile, so nothing updates. The emotion has to be present, not described. |
| Two-chair (a self-conflict or polarity; top-dog / under-dog) | Furnace-interrupt + precision — externalize an internal split so the suppressed side is contacted instead of avoided | 2 / 6 | With a fragile or dysregulated client it floods (floor 7 overwhelmed); titrate and build safety first, or it re-traumatizes. |
| Staying-with / “the safe emergency” (block the escape from a feeling) | Furnace-interrupt (blocks the avoid-move) — hold contact with the aversive experience until the interoceptive error lands | 6–7 | Held past the window of tolerance it becomes exposure without safety — precision on threat spikes and the valve slams. |
| Present-moment awareness (“what are you aware of now?”) | Precision — shift weight off the narrated story onto live bottom-up experience | 2 / 7 | Becomes an intellectual report about awareness (floor 1) rather than contact; the client narrates instead of feeling. Redirect to the body. |
| Exaggeration / repetition of a gesture or phrase | Precision — amplify a faint bottom-up signal until it crosses into awareness and becomes figure | 7 → 2 | Amplifies a signal the client can’t yet hold, or turns theatrical and disconnects from real affect. |
What PP adds. Gestalt’s founding intuition — the paradoxical theory of change, that you change by fully contacting what you are, not by forcing what you’re not — is precision language without the math. “Trying to change” is usually the furnace: a top-down wanted-prior dominating and manufacturing its own resistance. Dropping into full contact turns that precision down and lets the bottom-up present signal — the frozen emotion, the felt sense — finally carry weight and complete its interrupted update. That is why Gestalt reaches what talk can’t: it delivers error to floors 6–7 directly, in the body, by enactment. And it is why the power is also the hazard — a live floor-6 reactivation with the valve shut is re-traumatization, which is why titration and safety are not optional here.
Family Systems — the furnace in more than one head
Core logic in PP. The unit of prediction is the system, not the person. The symptom is held in place by a multi-person feedback loop (homeostasis): each member’s prior predicts the others’ behavior and acts to confirm it, so the family is a network of interlocking furnaces manufacturing each other’s evidence. The “identified patient” carries a symptom that is functional for the system’s equilibrium — which is exactly why individual insight so often can’t move it: the confirming evidence is supplied by other people, continuously, in real time. You have to change the loop.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Genogram (Bowen) | Read the multigenerational floor-4 priors — the templates transmitted across generations | 4 (across generations) | Stays diagnostic: a map with no change move, insight the system’s live furnace keeps overriding. |
| Enactment (structural, Minuchin) | Prediction error at system level — run the problem sequence live, interrupt it, introduce a new outcome in the room | 4 / 5 | The dominant member’s prior too high-precision to register the new sequence; or the therapist doesn’t actually block the old move, and the furnace completes. |
| Boundary restructuring (structural) | Precision — re-set the weighting across a boundary (enmeshment = precision too high, disengagement = too low) | 4 | Attempted without joining first — the homeostatic furnace expels the therapist before the new structure holds. |
| Detriangulation (Bowen) | Furnace-interrupt — remove the third party absorbing a dyad’s error so the dyad finally has to metabolize it | 4 | Pulling the triangle too fast floods the dyad (no containment); differentiation and safety come first. |
| Reframing / relabeling (strategic, Satir) | Precision + narrative — change the meaning the loop depends on so it loses its grip | 1 / 4 | A clever reframe the family doesn’t feel is just words; it has to fit their experience or the old meaning snaps back. |
| Prescribing the symptom / paradox (strategic, MRI) | Furnace-interrupt — make the symptom deliberate so it can no longer serve its involuntary system-confirming function | 5 | Used as a trick rather than from a real systemic formulation; it misfires and costs trust. Not for high-risk symptoms. |
What PP adds. It says precisely why individual work so often can’t move a family-maintained symptom: the confirming evidence is manufactured by other people in the system, continuously, so the client’s own furnace was never the whole engine — the network is. Family systems is the furnace scaled from one head to a relational web, and its interventions are furnace-interrupts and prediction errors delivered at the level of the loop — enactment, detriangulation, boundary-making, prescribing the symptom. It also reframes Bowen’s differentiation of self in a single line: the capacity to hold your own prior’s precision under the pressure of a high-precision shared emotional field — which is why the differentiated member can change the whole system simply by refusing to supply their scripted move.
Adlerian — testing the style of life
Core logic in PP. The “style of life” is a high-level generative model laid down in childhood — a single, self-authored strategy for how to belong and matter, organized around an imagined final goal (“fictional finalism”). “Private logic” is the set of rules that model runs on. Symptoms are safeguarding moves — the furnace protecting the style of life and the self-image it defends. The engine is teleological: a felt gap between where you are and the imagined goal (Adler’s “inferiority feeling” driving the “striving for significance”) pulls compensating action — which is active inference, behavior working to close the distance between a wanted-prior and the perceived present. Health, for Adler, is social interest: feeling connected to and useful to others — the sociometer reading well.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Early recollections (lifestyle assessment) | Read the narrative prior — early memories are selected to fit the present style of life, so they reveal the generative model and its private logic | 1 | Treated as literal history rather than a present-day projection; or gathered thoroughly and never converted into a change (assessment with no reorientation). |
| “The Question” (“what would be different if you were well?”) | Functional analysis — surfaces the symptom’s purpose, the furnace move it’s quietly running | 1 / stakes | Read as a gotcha, it shames; and a truly organic problem can answer “nothing” — don’t over-interpret it. |
| Encouragement | Precision / valve — raise a discouraged prior’s confidence just enough to risk a real test | conditions / 1 | Cheerleading untied to an achievable action becomes empty praise; encouragement is courage to act, not compliments. |
| “Acting as if” | Prediction error — act as the person you’d be if the belief were false, and let the disconfirming evidence land | 5 / 1 | Framed as pretending rather than an experiment with a written prediction; or pitched above the client’s courage, guaranteeing the failure. Start small. |
| “Spitting in the soup” | Furnace-interrupt — name the hidden payoff so the move can no longer run unseen | 1 / stakes | Delivered before the alliance can hold it, it’s an accusation; offer the payoff as a guess, not a verdict. |
| “Catching oneself” | Metacognition — the client learns to spot the furnace move as it begins | 2 | Stays intellectual — noticing with no replacement action; pair it with “acting as if.” |
What PP adds. Adler drew the engine before the vocabulary existed: his “style of life” is a generative model, his “fictional finalism” a high-level prior that organizes everything beneath it, his “safeguarding tendencies” are furnace moves, and his “social interest” is the sociometer. What the frame adds is the why under his best technique. “Acting as if” works not because pretending is powerful but because it is a prediction-error experiment — it puts the client in the one spot the style of life can be disconfirmed, doing the thing the private logic says they can’t, and lets the outcome count. And it explains “spitting in the soup” in a line: a furnace move survives only while it runs unseen, so naming the payoff out loud is what disables it — the same logic as Family Systems’ “prescribing the symptom,” scaled down to one person.
Postmodern (SFBT & Narrative) — reversing the filter
Core logic in PP. The postmodern approaches start from a different place: there is no single objective prior to correct, because the “problem” is itself a construction — a dominant story about a life — and change comes from building a different one. In these terms both Solution-Focused Brief Therapy and Narrative Therapy work floors 1 (the story) and 2 (your relation to it), and their signature move is a precision maneuver aimed at one furnace move in particular — the filter. A problem-saturated story throws away every moment that doesn’t fit; these approaches go hunting for exactly that discarded evidence and make it count.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Miracle question (SFBT) | Installs a vivid, high-precision future-prior (a goal-state) that reorganizes attention and action — and surfaces exceptions already present | 1 | A vivid future with no small first step built toward it is a wish, not a plan; anchor it to the next concrete sign. |
| Exception & coping questions (SFBT) | Precision — hunt the disconfirming evidence the “it’s always like this” filter discards | 1 / 2 | Mining for exceptions past a client in real crisis reads as dismissive; validate the suffering before hunting exceptions. |
| Scaling questions (SFBT) | Precision calibration + a small, achievable next-error (“what moves you one point?”) | 1 | Becomes a number game with no behavior attached; the point is the next testable step, not the score. |
| Externalizing (Narrative) | Defusion at the level of identity — drop precision on “I am the problem,” making it a thing outside the self | 1 / 2 | Externalizing a problem that needs ownership (behavior harming others) can dodge responsibility; separate the person from the problem, not from accountability. |
| Unique outcomes → re-authoring (Narrative) | Precision + narrative content — surface the moments the problem-story filtered out, then build the counter-story from them | 1 | A counter-story pasted over an unregulated concrete floor (6–7) stays words; thin re-authoring without felt evidence doesn’t hold. |
| Deconstruction / outsider-witness practices (Narrative) | Name the floor-8 discourse feeding the story; raise precision on the new story through social witnessing | 8 / 1 | Intellectual deconstruction that never reaches the felt sense; and witnessing needs a genuinely safe audience or it backfires. |
What PP adds. It names what these approaches are doing under the constructionist language: systematically reversing the filter. Where CBT argues with the belief’s content, SFBT and Narrative don’t argue at all — they redirect attention to the disconfirming evidence the story has been throwing away, which is a precision move, not a content one. And Narrative’s externalizing turns out to be ACT’s defusion performed at the level of identity: “the problem is the problem; the person is not the problem” is exactly “you are not your thought,” raised from a single cognition to the whole self-story. The caution the frame adds is the one it adds everywhere — a re-authored story that never reaches the concrete floors stays words; the new narrative has to be backed by felt, un-rigged evidence, or the old story snaps back.
Reality Therapy — active inference in plain language
Core logic in PP. Choice Theory holds that all behavior is chosen to meet five basic needs, and that each of us builds a “quality world” — an internal album of the specific people, things, and beliefs that would satisfy those needs. All behavior is an attempt to close the gap between that wanted-world and what we perceive we are getting. That is active inference stated without the math: the quality world is a set of high-precision goal-priors, and behavior is the action arm trying to make perception match them. Reality Therapy works the present, choice, and responsibility, and its whole engine is self-evaluation — the client judging whether what they are doing is actually getting them what they want.
| Technique | Lever pulled | Floor | When it fails |
|---|---|---|---|
| Exploring wants / the quality world (W) | Read the goal-priors — the specific images of what would meet the need | 1 | Wants-talk that never reaches evaluation or a plan; exploration with no conversion. |
| Evaluating current doing (D → E: “is what you’re doing getting you what you want?”) | Self-generated prediction error — the client, not the therapist, names the gap between wanted and achieved | 1 / 5 | If the therapist supplies the verdict, it fires the furnace and the client defends the behavior; the evaluation has to be theirs, like MI’s develop-discrepancy. |
| Making a plan (P: simple, achievable, measurable, immediate, committed) | Install a new action-script and a concrete next-error to run | 5 | A plan too big or vague guarantees failure and reconfirms “I can’t change”; keep it small and immediate. |
| Present focus + responsibility (“no excuses”) | Precision on agency — weight the controllable (acting, thinking) over the felt-helpless (feeling, physiology) | 2 / 5 | Slides into blame when the environment is genuinely the pathogen; “your choice” is the wrong sentence for a structural injury (the calibration gate governs which you’re facing). |
What PP adds. Glasser described the action half of the loop — behavior as the attempt to make perception match a wanted-model — decades before “active inference” had a name, and his central move is the sharpest example of a principle the frame states elsewhere: change lands harder when the person generates the discrepancy themselves. WDEP’s evaluation question is MI’s develop-discrepancy without the Rogerian wrapper — a self-generated prediction error, weighted higher because the client produced it. The one hazard the frame flags is the “responsibility / no excuses” stance: it is precision on agency, exactly right for a stuck-but-workable prior and exactly wrong when the pathogen is the environment — which is precisely what the calibration gate is there to sort.
This note re-describes what the modalities do; it does not claim their own theories are wrong, or that they reduce to this vocabulary. Convergence is at the level of practice, not theory — the point is corroboration, not takeover. Use the columns to cross-train and to troubleshoot a stall, never to argue a colleague out of their model. And hold the discipline the paper insists on: families, not brands. The rows name where each tradition’s signature move lands, not the boundary of the therapy — cross floors freely, and norm every move to the client’s culture (floor 8).
A conceptual and troubleshooting aid, not a treatment manual or an evidence review. Every mechanism claim here is a candidate re-description; some — EMDR especially — sit on open scientific debates. It presumes the gates in the Decision Aid — Locating the Floor have been cleared — acute risk, the reversed dial (psychosis/mania), and the calibration check — before any of it is used. Diagnosis, risk judgment, and standard of care are unchanged. See boundaries in A Unified Clinical Model of Psychotherapy Part 7.
*See also: A Unified Clinical Model of Psychotherapy · Case Formulation — One Page · Decision Aid — Locating the Floor · Decision Map — Locating the Floor · Clinical Applications — Status, Mattering, and Rank.*