FORWARD PATH — THE SYSTEM RUNS BLIND · IT DOES NOT KNOW WHAT WILL WORK BEFORE THE STRATEGY FIRES ① DETECT ② MATCH + FILTER ③ ASSEMBLE ④ FIRE signal body · environ. ZONE 1 Salience · Routing → threat detection → arousal level → urgency · valence → fast path fork output: shaped pressure high threat → SURVIVAL INTERRUPT subcortical · pre-model fires before Z2–Z3 no narrative · no concept shock trauma encodes here prior filter ZONE 2 Structural Matching → implicit / embodied priors → relational templates → survival predictions ⚠ THE PREMISE LAYER high-confidence priors filter incoming output: structured signal ZONE 3 Strategy Assembly → belief priors → concept selection → emotional construction → action policy selection output: emotional strategy ACTION strategy fires behavior · felt state ↓ feedback begins FEEDBACK LAYER ⑤ PROCESS ⑥ REFLECT ⑦ RECURSE ZONE 4 Reflective Compression → narrative construction → emotion labeling → meaning-making · analysis → causal story (post-hoc) why it worked · why I feel this way PROCESS FEEDBACK LOOP Prediction confirm? ✓ it worked ✗ disappointment Burner turn off? ✓ relief ✗ continued pressure ? ATTRIBUTION GAP — which variable caused the outcome? the system guesses → the guess becomes the reinforcement signal · the guess can be wrong strategy reinforced updates concepts becomes new prior feeds meaning-making SELF-REFERENTIAL LOOP Z4 narrative feeds back into Z2 premises — which then filter the next incoming signal. WHAT THE SYSTEM OPTIMIZES FOR Predictive accuracy — not objective accuracy. High-confidence priors filter data before it can challenge them. Without new experience at the right layer, the model confirms itself. HOW IT GETS STUCK Effective strategies prevent premise testing. The better the coping, the more reinforced the premise. The loop is self-maintaining — not because the person is failing, but because the system is working exactly as designed. shaped pressure structured signal emotional strategy upstream signal model · predictive compression system · mechanism