Memory and metamemory for social interactions: Evidence for a metamemory expectancy illusion

People do not always have accurate metacognitive awareness of the conditions that lead to good source memory. In Experiment1, participants studied words referring to bathroom and kitchen items that were either paired with an expected or unexpectedroom as the source. Participants provided judgments of item and source learning after each item–source pair. In line with previousstudies, participants incorrectly predicted their memory to be better for expected than for unexpected sources. Here, we show thatthis metamemory expectancy illusion generalizes to socially relevant stimuli. In Experiment2, participants played a prisoner’sdilemma game with trustworthy-looking and untrustworthy-looking partners who either cooperated or cheated. After each roundof the game, participants provided metamemory judgments about how well they were going to remember the partner’sfaceandbehavior. On average, participants predicted their source memory to be better for behaviors that were expected based on the facialappearances of the partners. This stands in contrast to the established finding that veridical source memory is better for unex-pected than expected information. Asking participants to provide metamemory judgments at encoding selectively enhancedsource memory for the expected information. These results are consistent with how schematic expectations affect source memoryand metamemory for nonsocial information, suggesting that both are governed by general rather than by domain-specificprinciples. Differences between experiments may be linked to the fact that people may have special beliefs about memory forsocial stimuli, such as the belief that cheaters are particularly memorable (Experiment3)