Training links may suppress copying

Online cultural copying may change when people expect a copy to shape audiences. This hypothesis asks whether a link to a recommender, a system choosing later content, can suppress copying. Participants would learn that copying triggers a later update to recommender training. The proposed test compares linked and disconnected training gates while matching exposure, content and copying opportunities. Delivery remains available, and copying is measured before later-audience feedback returns. D(t) is the difference in probability of a documented new human-produced descendant between disconnected and linked gates. More copying after disconnection would support gate-specific influence; reactance must also predict results when the training contingency moves to a new gate.