Case Formulation — One Page

A working frame from A Unified Clinical Model of Psychotherapy. Answer the five questions with the client — out loud, co-owned — and assemble the response. Not a protocol — a lens to hold alongside your training and your risk judgment.

The five questions

1 · WHERE does the stuck prediction live? (the floor) Which level is the problem actually on? Find it by the signs, not the story:

accurate insight that changes nothing → it's below where the words are reaching
the body answers first (heat, constriction, numbness, drop) → floors 6–7
the same pattern every time, across partners/jobs/decades → floor 4, relational template
it shows up with you (managing, testing, going compliant/opaque) → floor 4, in the room
the reaction too big for the occasion → follow the flare to the high-precision prior
the client's verbs: "I believe / I tell myself" (high) · "I always end up" (scripts) · "I just shut down / my chest closes" (low)

Floors, top→bottom: 1 Narrative · 2 Metacognition · 3 Beliefs · 4 Relational template · 5 Scripts/behavior · 6 Procedure/emotion · 7 Body · 8 Cultural substrate. Glass up top (words reach it); concrete down low (words don't).

2 · By what LEVER does it change? (precision) Assume the corrective information has already been heard and weighted at zero. The job is not to supply a new belief — it's to re-weight: lower confidence on the locked prior, raise it on the disconfirming evidence, so an error that's been arriving all along can finally count.

3 · Against what OBSTACLE? (the furnace) Which move is manufacturing the confirming evidence? Provoke · Avoid · Filter · Reinterpret · Withdraw effort. A corrective experience won't land while the client's own behavior is producing the opposite faster than you can. Name the move; plan how to interrupt it long enough for one unmanufactured error.

4 · Under what CONDITIONS? (the relationship) Safety and trust are what turn the dimmer up. Without them the signal stays weighted to nothing and no technique reaches its floor. The alliance isn't the backdrop — it's the mechanism. Sequence it first, then aim floor-matched technique through the space it opens.

5 · With what at STAKE? (mattering vs. rank) Which social gauge is injured?

Mattering (sociometer — wanted, needed, included) — local, concrete, reachable from your office. Joiner's burdensomeness & belonging both live here. Usually the lethal one.
Rank (hierometer — standing, position, comparison) — you can't hand it over in 50 minutes; don't pretend it doesn't matter. Recalibrate the reference class instead of the rank.

The intervention engine: prediction-error experiment

Step 0 — is the prior false? Some predictions are accurate reads of a punishing environment (contemptuous partner, punishing workplace, real discrimination). If the world will confirm the fear, fix the field first — safety, boundaries, advocacy — and don't run disclosure into it. Then target-select: ladder first — safest person, not the most important.

Reframe the risky act as an experiment, never a demand to "be vulnerable" (heard as a demand to capitulate). "I bet you're right — let's get the data."

Name the prediction — "if I tell someone I'm struggling, they'll think less of me."
Rate confidence + write it down, before — the number and the writing make the gap undeniable; memory rewrites an unrecorded prediction to fit the result.
Run the smallest real test — one named person, one true thing, once.
Compare, out loud — the gap between predicted and actual is the update. Protect this step; a rushed session drops it.
Ask how many times it'd take to revise the rule — turns one result into a standing experiment.

Starter kit (aim each at its floor): disclosure — one true thing · the ask · escalation ladder · delayed report · text (relational) · irreplaceability inventory · unglamorous favor · reference-class swap (mattering/rank) · block the manufactured evidence (furnace) · build-with-them · dead-air · satiation check (appetitive wall) · exposure as expectancy-violation (fear).

Which wall? Before intervening, log the antecedent: escape fires after aversive moments → defensive, needs disconfirmation (epistemic). Escape fires out of neutral/good states (an idea landed) → appetitive, needs capture-the-impulse + a 24-hr delay (behavioral) — there's no false belief to disprove. And for a builder, don't let the experiment program become the next build — index card, not a dashboard.

The three disciplines

Desert move: separate the reading from the verdict. The rank/loss may be real; "therefore I'm contemptible and deserve it" is the false part. Strip the desert → same fact, different affect (forward agency stays — the steering wheel survives the verdict).
Norm to culture: restraint ≠ pathology in traditions where it's ordinary dignity; choose tests — and markers of "closeness" — the client's world actually uses.
Don't prescribe your own cure. If striving saved you, you'll be tempted to hand a stuck client the climb. He tried it; the world didn't cooperate; that's why he's angry. Test your prior before you prescribe it.