Do not open with housekeeping. Open with the hook on the next slide and get to the housekeeping in the last two minutes, when attention is gone anyway.

Do not explain the exercise. Put it up, say "read both, decide which", and go quiet. The silence is the instruction. Thirty seconds is enough.

Then take a show of hands for A, then for B. Do not say who is right. Say: "Hold that. We are going to come back to it in forty minutes and I want you to notice whether you change your mind."

Ask what the friend did. Answers will be: reassured, normalised, offered perspective. All correct, and all things therapists also do.

Then ask what the friend did with "I still think it." Nothing. It went past. That is the observation to bank — not for a verdict yet, just bank it.

Now ask the same question: what did this therapist do? The room will say "asked questions", which is not the answer. Push: the therapist went *towards* the thing the friend went past.

The load-bearing move is "What would it mean about you". It converts a symptom into a hypothesis about function. That is a theory of change operating, whether or not the therapist could name it.

The number is not the interesting part. The interesting part is that nobody can agree on it, and the reason nobody can agree is that there is no accepted criterion for when two therapies are the same therapy.

That is the actual problem this course solves. Not "learn 45 therapies" — learn one way of interrogating any therapy, including ones invented after you graduate.

They must physically write it. On paper, in a notebook, in a phone — but written and dated, because they will compare it to a second attempt at the end of the session and the comparison only works if the first one is fixed.

Take two out loud. Do not evaluate them. Write both on the board verbatim; you will dismantle them in fifteen minutes and it is much better if the examples came from the room.

Read it out once, slowly. Then say: "That is a mouthful and every clause in it is load-bearing. If I delete any one of them, something we want to exclude walks in."

Then go clause by clause on the next slide. Do not paraphrase the definition first — the paraphrase is the thing that loses the clauses.

Work down the rows and make them supply the counterexample before you show it. "What gets in if I delete *informed and intentional*?" Someone will say a good friend. Correct.

Spend the time on the last row. The others are housekeeping; that one is the moral spine of the definition and of the profession.

This is the clause that distinguishes psychotherapy from social control, and it is not academic. Practices designed to change sexual orientation or gender identity fail this clause: the direction of change is set by somebody other than the person in the chair.

They are also ineffective and associated with harm, and are prohibited or restricted in a growing number of jurisdictions. But the definitional point stands independently of the evidence and independently of the law: even if such a practice worked, it would not be psychotherapy, because the goal was not the patient’s.

The transferable lesson: a definition is not bookkeeping. Where you draw this line determines what you are willing to be paid to do.

The move here is to resist the students’ instinct to rank these. Peer support is not lesser psychotherapy; it is a different thing that works through partly different mechanisms.

Ask: which of the three would the definition admit, if we were strict? Coaching gets in on method and out on population. That the definition cannot cleanly exclude coaching is a genuine weakness, and admitting it now buys you credibility for the rest of the course.

Run it as peer instruction, properly

1. 1Everyone votes silently, hands up or app. Record the split on the board.
2. 2Find someone who disagrees with you and convince them. Two minutes, loud.
3. 3Vote again. Show both splits.

The point is that the first two options are *both* correct, and the vote will move. What moves it is students discovering that "interpersonal" is ambiguous — which is a discovery about the definition, arrived at by argument rather than by being told.

Retrieval, not review. They must produce it from memory; looking at their notes defeats the entire purpose and is worth saying out loud.

Then: "Compare it to what you wrote at the start. If it has not changed, either you were unusually good or you were not listening." That comparison is the session’s evidence that something happened.

Open by asking what the definition was for. Answer: it told us what is in scope. It told us nothing about how to compare the four hundred things now in scope. That is this half.

Be careful not to be cynical here. Almost nobody in this chain is acting in bad faith. Founders genuinely believe their synthesis is distinct, and allegiance operates through a hundred small honest decisions rather than through deceit.

But the structural incentive is real: a system that described itself as "mostly the same as the last one, with a different emphasis on homework" would not attract trainees.

Two questions, asked of everything, in the same words every time: which processes of change does it use, and at what level of content does it aim them?

This is what makes forty-five systems comparable rather than forty-five things to memorise separately. It is the single most useful idea in the course, and it is worth saying so.

Do not lecture all ten. Take three and make them concrete, then tell the room the rest are in the app coded against every system, which is where they will actually learn them.

• Consciousness raising: you come to know something about yourself you did not know. Psychoanalysis lives here.
• Counterconditioning: the old cue stops producing the old response, because a new one has been trained onto it. Exposure lives here.
• Helping relationship: the relationship itself is the agent, not the vehicle. Person-centred lives here.

The caption is the point and students will resist it, because the stack looks like a hierarchy of profundity and our culture rewards depth talk.

Give them the counterexample immediately: for a specific phobia, the level that works is the top one, and going to the bottom one is not deeper insight, it is a longer, more expensive failure.

This grid is the assessment structure for the whole course and most of the assignments. Say that explicitly; students work harder on a framework they know they will be examined through.

Emphasise row 3. If you can only ask one question of a therapy, ask what it claims actually produces change, then ask what would happen if that claim were false.

This is the highest-yield slide in the unit for later assignments. Most weak student writing collapses these four, and the collapse is invisible to the writer.

Test it live: give them "the patient has maladaptive schemas". Which of the four is that? Pathology. Now: what is the theory of change that follows, and what is the technique? They should notice that the label alone commits you to nothing.

The pedagogical purpose is to stop them concluding "all therapies are secretly the same", which is the fashionable overcorrection and is false.

The honest position: the gap between exposure and person-centred is real and large. The gap between two manualised cognitive therapies frequently is not. Which pair you pick determines the answer, and that is why "are the differences real?" has no single answer.

Give both sides at full strength. Substantive: mechanisms differ, and for some conditions the mechanism-matched treatment demonstrably outperforms. Semantic: bona fide treatments delivered by people who believe in them produce broadly comparable outcomes across most presentations.

Then refuse to resolve it. Say: "I have a view. I am not going to tell you what it is until Unit 17, because if I tell you now you will spend nine weeks agreeing with me instead of looking."

Thirty seconds alone before they turn to each other. Without the silent think, the confident student in each pair sets the answer and the other one agrees.

The best answer available: they are different treatments if the mechanisms predict different things — different patients responding, different relapse patterns, different failures. If they predict nothing different, the distinction is not doing scientific work.

Cold call rather than volunteers. Volunteers tell you what the strongest student knows, which is not the quantity you need.

Anything the room cannot produce here is what the next session opens with.

The third question is the one that opens Unit 3, and it is deliberately the one they have least equipment for. Leave it hanging.

Sell the self-survey properly or it will be done badly. The line that works: "You already have a theory of how people change. You absorbed it from your family and from whoever helped you when you needed it. Writing it down now is the only way to find out later whether you chose your approach or whether it chose you."

Tell them it will be sealed and returned in Unit 20. That single promise measurably improves the honesty of the answers.

Leave it up while they pack. Nobody reads it, and it should still be there.