Prediction error parameters for visuospatial interference: Minimal prediction errors leave positive declarative memory intact following visuospatial word search interference

Abstract

Prediction error (PE) is an important boundary condition for memory reconsolidation; however, while it is clear that a prediction error is necessary for reactivation, the parameters for what constitutes a prediction error are unclear. In the present study, participants watched a video with a positive emotional valence, followed by a baseline free recall test. One day later, participants completed one of three prediction error conditions (NoPE, 1PE, 2PE), followed by a visuospatial interference task (i.e., complex word search) or a nonvisuospatial multiplication control task. The NoPE condition viewed a photo and short summary of the video presented in a congruent modality (i.e., digitally on a computer monitor). The 1PE condition viewed the same photo and summary but received this information in an incongruent modality (i.e., on a piece of paper). The 2PE condition viewed the same photo and summary in an incongruent modality (i.e., on a piece of paper), plus one new fact added to the summary. Memory testing one week later showed a significant drop in free recall, but not recognition, for all groups from the first to the second free recall test; however, no differences were observed between groups. These results indicate that, in the positive film paradigm, minimal prediction errors (i.e., one or two changes) are not sufficient to destabilize memory. Given the importance of prediction error for applications of reconsolidation to the treatment of memory-based psychological disorders and drug addiction, further examination of the parameters for prediction error is warranted.