Goal: make an earlier feature prediction and compare it with a later approved observation only when both actually occur. Proposed codes: AC9SFI01, AC9SFI02 (approved direct live view only), AC9SFI04 (genuine prior/result only), AC9SFU01. Prepare: new approved distinct living/photo source if available, gate, two-pair log, Card 148.
- Gate · 3 min. Check approval, distinct identity, rights, date, access and no disturbance. If unavailable, label N model exercise and REAL OBSERVATION NOT AVAILABLE.
- Prior words · 4 min. Before any new genuine view, child predicts a visible external feature or selects NO PREDICTION; record exact words and timing.
- View/model · 5 min. Reveal only the approved real view and source label, or read a different N drawing as an explicitly fictional paper reveal.
- Record · 7 min. Card 148 routes write exact later visible feature, NOT VISIBLE or UNCLEAR; record LIVE/PHOTO versus ADULT REPORT. Do not copy a model feature into actual log.
- Compare · 4 min. Compare only a true earlier prediction with a true later observed feature; otherwise write CANNOT COMPARE. “My guess was right” without a view is not evidence.
- Exit · 2 min. Save order, source and one bounded next question.