Model Corrections Provoke Deliberate Errors

Why can an accurate model correction make people retell a story less faithfully? This hypothesis asks whether overriding an earlier right to approve changes can trigger deliberate distortion. In matched fictional story histories, participants or neutral editors would hold final approval. Everyone would receive the same source-checked correction. One path would respect that approval; another would override it. Researchers would compare public retellings with private answers about the source at the same story version. If public accounts drift only after an override while private answers stay accurate, that pattern would support the proposed mechanism. It would still have to predict better than learning, ownership, or ordinary reconstruction.