What characteristics make self-generated memory cues effective over time?

From writing to-do lists to creating mnemonic devices in school, people frequently generate cues to help them remember information. Creating memory cues is a vital aspect of metacognition and allows learners to somewhat control their retrieval circumstances. Across three experiments, we tested the extent to which self-generated memory cues fail at long retention intervals because they are based in fleeting mental states. Participants studied target words and generated mnemonic cues for themselves or for others. Cues intended for others showed greater cue-to-target associative strength, were less distinctive, and were less idiosyncratic (more common) than cues intended for oneself. However, the effectiveness of the cues in supporting recall did not differ by intended recipient at medium (∼3 days) or long (∼1 year) retention intervals. In the third experiment, we directly tested the stability of self-generated cues for oneself (compared to cues for others, descriptions of the target, and focused descriptions) by asking participants to generate cues twice for the same targets across a delay of 3 weeks. Cues intended for others were more stable than all other cues, but the stability of the cues did not affect long term retention. Implications for effective cue generation are discussed.