Relative contributions of semantic and phonological associates to over-additive false recall in hybrid DRM lists

Two experiments explored false recall of unstudied critical items (e.g., chair) following the presentation of 16 semantic associates to the critical word (e.g., sit, desk), 16 phonological associates to the critical word (e.g., cheer, hair), and every composition of hybrid list in between (e.g., 14 semantic and 2 phonological associates). Results replicated the over-additive pattern of critical false recall from hybrid lists relative to pure lists found by Watson, Balota, and Roediger (2003) and clarified the form of the false recall function across varying degrees of hybridization. Both experiments showed that including just one or two of the other type of associate in an otherwise pure list led to a considerable increase in false recall. A within-subjects design (Experiment 1) suggested that after this initial rapid increase, false recall continued to increase gradually to an apex at the balanced hybrid list composition, whereas a between-subjects design (Experiment 2) showed that false recall plateaued after the initial rapid increase and that the overall shape of the function is a ziggurat. Furthermore, the function is roughly symmetrical; semantic and phonological associates appear to make equivalent contributions to over-additive false recall from hybrid lists. The results provide constraints on theoretical accounts of DRM false memories, and can be accommodated by a modified activation/monitoring framework.