Gamma prompts — REE for Psychiatrists

Generation prompts for the six-part REE for Psychiatrists deck. Each block is a standalone prompt that produces one ~10-card chapter. Paste one block at a time into Gamma via Create new → Paste in text (paste mode, not “Generate from a prompt”), choose one card per divider, then merge the six decks in Google Slides (File → Import slides, keep original theme).

Each prompt opens with a DO NOT guard banning the generic REE intro slides — without it, Gamma regenerates the Part 1 framing for every chapter. Shared instruction for all six: audience = consultant psychiatrists, no ML background; grand-rounds tone; explain each term plainly; one idea per card (headline + 3–5 bullets); honesty guardrail — REE is a research-stage specification, not a deployed AI, clinical tool, or validated disease model.

Created 2026-06-12. The merged deck (corrected: Part 2 ladder numbered from 0; Part 5 UK spelling) is the canonical copy reproduced on the REE for Psychiatrists page.

---

Part 1 — Why build a mind from first principles?

Standalone 10-card deck: Part 1 of a 6-part series for psychiatrists.
Card 1 TITLE "REE — the Reflective-Ethical Engine" / "A series for psychiatrists · Part 1 of 6" + honesty guardrail.
Card 2 The framing question — minimal commitments without which thinking is incoherent; developmental psychiatry / phenomenology / REE ask it (computationally, testably).
Card 3 What REE is, in one breath — world-model; planning/commitment system; control layer.
Card 4 What REE is NOT — not a moral rule-engine; not reward-maximisation-with-guardrails; not a chatbot/consciousness claim; not a finished product.
Card 5 Why a psychiatrist should care — over-precise belief→delusion-like lock; trajectory collapse→depressive narrowing; residue overload→guilt/moral injury; other-model collapse→instrumental relating.
Card 6 The central commitment — moral continuity; ethics cannot be optimised to zero; residue as conscience.
Card 7 Why residue is necessary — forecloses futures; harm untraceable; minimised unnecessary harm + honest accounting.
Card 8 Method preview — registered claims, stress-tested, governed ledger; not features-ship-when-tests-pass.
Card 9 Series map — the six parts.
Card 10 Closing frame — "We are uncertain minds, together in a shared world, capable of love…"; derived not asserted.

Part 2 — From Axioms to Ethics

Standalone 10-card deck: Part 2 of 6. DO NOT include a framing-question / what-REE-is / what-REE-is-not / why-care / series-map / generic-closing slide (those are Part 1). Theme: how a handful of commitments force an ethics.
Card 1 TITLE "REE — Part 2: From Axioms to Ethics" / Part 2 of 6.
Card 2 Articles of faith, not design choices — irreducible axioms; infant-mind analogy.
Card 3 Dependency ladder, layers 0–3 (number bullets 0,1,2,3): 0 I think therefore I am; 1 existence has value; 2 certainty unavailable yet action required; 3 agency is bidirectional (power + vulnerability).
Card 4 First derivations — D1 self-preservation; D2 model refinement (derived, not added).
Card 5 Social layers 4–7 — others exist; shared responsibility; love as mechanism; language (attachment, mentalising, repair).
Card 6 Love as mechanism, not sentiment — long-horizon coherence bias; secure attachment vs instrumental relating.
Card 7 The pivot: ethics is derived (Layer 8) — follows necessarily from 0–7 + D1/D2; "REE is the machinery, not the morality".
Card 8 Derived ethical objectives — preserve minds; future options; reduce suffering; shared joy; corrigibility; truth-seeking; love; honest communication.
Card 9 Three functional constraints — rapid prediction; temporal depth; constrained commitment (become the architecture in Part 3).
Card 10 Closing — the full ten-layer ladder Self→…→Ethics→REE.

Part 3 — The Architecture, Mapped to the Brain

Standalone 10-card deck: Part 3 of 6. DO NOT include the generic intro slides. HONESTY on card 1 + card 10: brain-region mappings are theoretical homologies, NOT biological fidelity.
Card 1 TITLE / Part 3 of 6 · E1, E2, E3, control plane, hippocampus + homology guardrail.
Card 2 REE is a prediction-and-error machine — predict / compare / correct; three constraints become four components.
Card 3 E1 persistent predictive substrate — slow world-model; addressable associative memory; cerebral cortex / parietal; "what remains when attention drops".
Card 4 E2 fast transition model — short-horizon "what next if I do X"; trains on motor-sensory error; cerebellar forward-model intuition.
Card 5 E3 trajectory selection & commitment — evaluates futures, gates irreversible attributable action; prefrontal + hippocampal rollout.
Card 6 Hippocampal rollout — proposes multi-step futures across E1; prospection (impaired in depression, over-rolling in anxiety).
Card 7 Control plane — precision + mode; perception↔default-mode; aberrant precision in psychosis.
Card 8 Dark until ready — E3 present but dark until E1/E2 mature; moral capacity needs scaffolding.
Card 9 The circuit assembled — sensory→E1/E2→hippocampal rollout→E3 gate→action→learning/residue; control plane over all.
Card 10 Closing (honesty) — shared structural vocabulary, not a brain model. Tee up Part 4.

Part 4 — Affect, Drive, and Moral Residue

Standalone 10-card deck: Part 4 of 6. DO NOT include the generic intro slides. Lean into phenomenology. Theme: affect/drive are control signals that shape which futures get selected.
Card 1 TITLE / Part 4 of 6 · why a principled mind must have something like feeling.
Card 2 Affect as control signal, not add-on — computed quantities biasing selection; affect as information.
Card 3 Homeostatic drive — self-state sensing (damage/depletion/threat); interoception; drive states.
Card 4 Self-sensing and harm-sensing — "this is happening to me" / "this is damage"; prerequisite for ethics.
Card 5 Empathy as model-reuse — self-model predicts others; "what hurts me"→"what hurts you"; mentalising.
Card 6 Moral residue (core construct) — persistent geometric cost, even for justified acts; guilt/conscience/moral injury.
Card 7 Residue must be balanced by repair — offline sleep-like replay reduces spurious residue, preserves genuine cost.
Card 8 Candidate-differentiation principle — affect must differ across options; flat landscape carves no behaviour; flattened affect.
Card 9 Why ethics needs affect — without residue/drive ethics collapses to loophole-exploiting optimisation.
Card 10 Closing — conscience as architecture (standing cost function). Tee up Part 5.

Part 5 — Computational Psychopathology

Standalone 10-card deck: Part 5 of 6 (the chapter the audience came for). DO NOT include the generic intro slides. HONESTY on card 2 + card 10: computational failure modes, shared vocabulary + hypotheses, NOT a diagnostic system; REE is not a validated disease model. For each failure mode give mechanism / expected behaviour / clinical rhyme. Use UK spelling (behaviour, optimise, generalise, programme).
Card 1 TITLE "When the Architecture Mis-Tunes — Computational Psychopathology" / Part 5 of 6.
Card 2 Honest disclaimer — "which computational parameter, mis-set, would produce this picture?".
Card 3 FM1 precision misrouting — over-precise context/stimulus/regime; aberrant-salience psychosis.
Card 4 FM2 trajectory-space collapse — depressive pruning of imagined futures; hopelessness, anhedonia.
Card 5 FM3 residue overload — paralysis; pathological guilt, OCD-like inhibition.
Card 6 FM4 spurious residue — false dents; contamination fears, magical thinking, anxiety generalisation.
Card 7 FM5 moral amnesia — residue disabled; callous-unemotional / antisocial.
Card 8 FM6 other-model collapse — others computationally invisible; empathy failure / dissociation.
Card 9 FM7 control-plane regime mis-tuning — hypervigilance, dissociation, rumination, mania; state-dependent switching.
Card 10 The payoff — parameter mis-set → predicted picture → testable hypothesis; not a nosology.

Part 6 — A Self-Governing Research Programme

Standalone 10-card deck: Part 6 of 6. DO NOT include the generic intro slides. Draw the EBM / pre-registration analogy. Theme: REE is built like a governed body of evidence, not features-ship-when-tests-pass.
Card 1 TITLE "How REE Develops — A Self-Governing Research Programme" / Part 6 of 6.
Card 2 The problem this solves — untested assumptions; pre-registered protocol vs post-hoc storytelling.
Card 3 The epistemic ledger — numbered claims with status + confidence + audit trail.
Card 4 The closed loop — idea→claim→experiment→evidence→governance→architecture→new experiments.
Card 5 Experiments as test bench — substrate; pre-stated pass/fail; a FAIL is information.
Card 6 Governance — contradiction flagged, confidence dropped, conflicts surfaced; data-monitoring committee for a theory.
Card 7 Discipline rails — gates; supersession; independent review; make self-deception expensive.
Card 8 Where it is now, honestly — early single-agent substrate; social/developmental claims need multi-agent substrate (future work).
Card 9 Why it matters to psychiatry — generative, mechanistic, falsifiable; governance models epistemic honesty.
Card 10 Closing statement — "We are uncertain minds…"; questions + collaboration welcome.