Predicting others' memory performance: The accuracy and bases of social metacognition

- Learners judged the effectiveness of mnemonic cues for others.
- Learners predicted others' recall more accurately than chance.
- Learners predicted others' recall less accurately than their own.
- Learners' predictions were not harmed by own idiosyncratic cue generation.
- Learners' predictions were hampered by insufficient information about others.

Successful teaching, effective advertising, and happy interpersonal relationships depend upon accurately anticipating what others will remember. Across three experiments, we tested how precisely subjects judged the mnemonic effectiveness of cues for supporting other subjects' episodic memories. Some subjects generated cue-target word pairs and made judgments of learning (JOLs) for these word pairs while other subjects studied the pairs and made JOLs. Across all three experiments, subjects' JOLs for others were more accurate than chance, but less accurate than subjects' JOLs for themselves. Further, JOLs for others were similarly accurate across cues that subjects generated for others and cues that subjects read but did not themselves generate. Idiosyncratic cue generation processes impacted subjects' JOLs for others; however, this bias was not the primary reason for the inaccuracy of JOLs for others. Rather, our results suggest that the accuracy of judgments about others' memories suffers because people do not have access to the personal idiosyncrasies of others' encoding and processing.