Disentangling the Effects of Backward/Forward Associative Strength and Theme Identifiability in False Memory

Abstract

Background: False memory has been extensively studied using the Deese/Roediger-McDermott paradigm. Despite the robustness of the effect, there is wide variability in the results, which is not fully understood. Method: Three experiments independently examined the role of backward associative strength (BAS), forward associative strength (FAS), and theme identifiability (ID) on false memories. In Experiment 1, lists varied in BAS while controlling FAS and ID. In Experiment 2, FAS was manipulated while BAS and ID were controlled. Finally, in Experiment 3, lists varied in ID while controlling BAS and FAS. Data was analyzed using both frequentist and Bayesian analyses. Results: We found false memories in all three experiments. Specifically, false recognition was higher in high-BAS than in low-BAS lists in Experiment 1. In Experiment 2, false recognition was higher in high-FAS than in low-FAS lists. In Experiment 3, false recognition was lower in high-ID than in low-ID lists. Conclusions: These findings suggest that both BAS and FAS—variables that promote error-inflating processes—and ID—which promotes error-editing processes—contribute independently to the production of false memories. Splitting apart the role of these variables helps to understand the variability of false memories and to extrapolate DRM tasks to explore other cognitive domains.