Conflicting revisions erode story memories

This proposed hypothesis asks whether conflicting revisions can weaken connected story memories after correct text returns. Researchers would give returning participants matched fictional narratives. They would keep final text, incompatible propositions, task duration and output length constant, then concentrate corrections in one connected neighborhood or scatter them across links. Causal bindings are remembered links among events, roles and exceptions. It predicts accelerating errors in nearby causal links after restoration for that participant; a fresh participant should reset. Support requires calibrated binding accessibility to predict held-out loads or graph structures without refitting. Reconstruction and competing-information interference must fail on those trajectories.