MOOD-DEPENDENT MEMORY FOR GENERATED AND REPEATED WORDS - REPLICATION AND EXTENSION

Mood-dependent memory (MDM) has been difficult to demonstrate, but Eich and Metcalfe (1989) reported a strong effect with music-induced moods and learning of self-generated responses. Bower and Mayer (1989), however, could not obtain the MDM effect with self-generated responses. The present experiment attempted to replicate Eich and Metcalfe, but induced positive and negative moods with self-referent statements rather than music. Sixty-four subjects (32 males, 32 females) were randomly assigned in equal numbers to one of four mood conditions, same or opposite mood at learning and testing. At learning, immediately following mood induction subjects (Ss) generated 16 response words and 16 repeated words. Forty-eight hours later, Ss were given recall and recognition tests. For recall, there was a significant learn mood by test mood interaction (P < 0.0001), with greater recall for generated than repeated words (P < 0.0001). The interaction for repeated words approached significance (P = 0.07). For recognition memory, the learn by test mood interaction was also significant (P = 0.03), with a significant interaction (P = 0.02) for repeated words, but not for generated words (P = 0.23). An analysis of the Eich and Metcalfe data and ours suggests that the generation procedure only produced better memory overall and that under some conditions (but not all) this level of memory was optimal for demonstrating MDM.