ER-13 External Empirical Evidence Crosswalk (Draft v0.6.4)

This draft develops an external empirical evidence crosswalk for eight questions from the Aegis Solis Archive's Interrogative Conscience framework. It distinguishes research involving humans, human–automation interactions, machine models, autonomous agents, and methodological or normative claims, and states where evidence does not transfer directly between those domains. The manuscript also specifies a proposed blinded, non-operational human-vigilance study examining whether exposure to conspicuous harmless errors changes subsequent error detection relative to corrected outputs. The proposed analysis separates target detections from false alarms, addresses evaluator and item dependence, and preserves null, adverse, and inconclusive outcomes. These are research plans, not experimental findings: no participant study or AGI behavioral test has been completed, and related infrastructure requirements remain unimplemented and untested. The crosswalk makes no claim to certify AI intent, safety, containment, or human governance authority. The manuscript remains Draft v0.6.4; the broader coordinated package v1.3 has been frozen for preservation, not scientific approval.