Visuospatial word search task only effective at disrupting declarative memory when prediction error is present during retrieval.

Prediction error is recognized as a necessary boundary condition for memory reactivation and reconsolidation. Furthermore, behavioral manipulations (e.g., visuospatial interference tasks, like Tetris) have been shown to be effective at disrupting reactivated memory. In the present study, participants created a memory with positive valence by viewing an uplifting video of a young boy who built an arcade out of cardboard boxes. Two weeks later, memory for the video was reactivated with a prediction error (i.e., new information was added) or without a prediction error (i.e., no new information was added). Following memory reactivation, participants completed a novel visuospatial interference task (i.e., a complex word search) or a control task (i.e., sitting quietly). One week following reactivation, participants completed two memory tests (i.e., free recall and recognition). Results showed that the visuospatial interference task was effective in reducing free recall, but only in the group that received a prediction error during retrieval. No other significant differences were observed, including on the recognition test. These results expand the literature on reconsolidation by showing that destabilization of memory with a positive emotional valence requires a prediction error and that a novel visuospatial interference task (i.e., complex word search) is an effective behavioral manipulation for decreasing free recall.