Gesture reinstatement tests role preservation

This hypothesis asks whether gestures preserve who did what when people retell a story with an artificial intelligence model. People would learn gestures for roles, then reproduce the same text while repeating the mapping or swapping it. Matching gestures should preserve causal roles, while swapping should create more reversals. Conditions must match practice, delay, workload and exposure, with no-gesture and meaningless-movement controls. Established multimodal memory must be calibrated: if it predicts this pattern, the proposed mechanism loses explanatory priority.