Large-billed crows (Corvus macrorhynchos) have retrospective but not prospective metamemory.

Evidence of metamemory, the ability to monitor one's own memory, has been obtained in some primates, but it appears to be weaker in other species. In this study, we examined whether crows flexibly modulate their behavior by monitoring the strength of memory trace in a delayed matching-to-sample task using two paradigms. First, crows performing a memory test were given an escape option to decline taking the test (prospective metamemory). Second, crows were given the escape option as a "not confident" report after completing the test (retrospective metamemory). Accurate memory performance yielded a reward with a higher probability, whereas inaccurate memory performance resulted in no such recompense. The escape option yielded a reward with a lower probability. In the prospective metamemory test, crows escaped the memory test more frequently with longer delay intervals than they did with shorter delay intervals but no more frequently in the sample-omission than the sample-present trials, indicating that the crows decided to take the test or decline it by using the delay interval as a cue. In contrast, in the retrospective metamemory test, the crows escaped the memory test more frequently when their memory-test response was incorrect than correct and more frequently in the sample-omission than the sample-present trials, indicating that the crows recognized confidence regarding their choice in the memory test and utilized the escape option to maximize reward probability. Although these results suggest that crows retrospectively monitor the strength of memory trace, their prospective metamemory ability has not yet been confirmed in the present paradigm.