Metacognitive Monitoring and Control Processes Underlying Generation-Potentiated Learning

The generation effect refers to the finding that self-generated material (e.g. h_rse) is better remembered than material that is passively read (e.g. horse). Using a two-block mixed-list paradigm, we replicate previous research showing that the recall advantage for generated words in the first block is attenuated in the second block (Experiment 1), suggesting that participants change processing strategies for read words in the second block (i.e., generation-potentiated learning). To assess metacognitive awareness of this effect, we measured immediate versus delayed judgments of learning (Experiment 2), restudy decisions (Experiments 3A and 3B), and study duration (Experiments 4A and 4B). The results showed a dissociable pattern, such that immediate measures were influenced by ease of processing (i.e., read items were given higher judgments of learning, chosen to restudy less often, and studied for less time) while delayed measures were influenced by ease of retrieval (i.e., generated items were given higher JOLs, chosen to restudy less often, and studied for less time). However, additional study of read words in the delayed condition did not eliminate the generation advantage in the first block, suggesting a labor-in-vain. These findings suggest that control decisions are more effective when based on information diagnostic of subsequent memory (i.e., ease of retrieval), but that generation-potentiated learning may not occur until after the first test.