False memory for faces is produced by the Deese-Roediger-McDermott paradigm based on the morphological characteristics of a list

Previous studies on false memory of non-semantic stimuli reported that false memories were produced for items that were unstudied but that were similar to the studied items. However, it remains unclear what representations of false memory are constructed by studied items and how the representations of studied items contribute to the occurrence of false memory. Therefore, we examined whether false memory is produced in the Deese-Roediger-McDermott (DRM) paradigm for non-semantic stimuli (i.e., facial lists), which was divided according to the morphological characteristics quantified by principal component analysis, and whether the false memory rate could be quantitatively predicted by the morphological characteristics of the studied items with a regression analysis. Thirty-five participants memorized five faces and performed a recognition memory test, including an unstudied lure item whose morphological characteristics were approximately the average of the studied items. The results showed that the lure items were more likely to be falsely remembered than other unstudied items whose morphological characteristics were not the average of the studied items, indicating that false memory of non-semantic stimuli was produced in the DRM paradigm owing to the morphological characteristics of the studied items. In contrast, the false memory rate could not be quantitatively explained by each kind of morphological characteristic. This indicates that false memory was not caused by the particular morphological characteristic of the studied items. Moreover, these results implicate that false memory of non-semantic stimuli in the DRM paradigm was provoked by gist representation of the studied items.