Attempted Retrieval Benefits are Limited in Realistic Learning Settings, Unless Used for Prior Knowledge-Based Personalization

In typical retrieval practice, the expected answer is given along with the cue when learners first encounter a new study item (initial passive study). Withholding the answer on this first encounter and giving learners only the cue (initial attempted retrieval) turns this first encounter into a retrieval opportunity. Previous lab studies have found that initial attempted retrieval can benefit learning. However, these studies typically used well-controlled yet designs deviating from realistic learning conditions. We here examine whether and under which realistic conditions (i.e., items are repeated multiple times, feedback is provided and learners have varying degrees of prior knowledge on the topic) initial attempted retrieval promotes learning. In four experiments (total N = 210), we confirm that there is a benefit of attempted retrieval on subsequent recall, even if the retrieval attempt is unsuccessful. We also show that this benefit quickly diminishes when items are practiced multiple times, limiting its practical significance in realistic learning settings. However, we also show that attempted retrieval does provide a method for identifying prior knowledge, which can be used to increase learning efficiency in multiple-repetition retrieval practice. These results can inform the application of attempted retrieval in the classroom.