The mismeasure of memory: when retrieval fluency is misleading as a metamnemonic index.

The experiments address the degree to which retrieval fluency--the case with which information is accessed from long-term memory--guides and occasionally misleads metamnemonic judgments. In each of 3 experiments, participants' predictions of their own future recall performance were examined under conditions in which probability or speed of retrieval at one time or on one task is known to be negatively related to retrieval probability on a later task. Participants' predictions reflected retrieval fluency on the initial task in each case, which led to striking mismatches between their predicted and actual performance on the later tasks. The results suggest that retrieval fluency is a potent but not necessarily reliable source of information for metacognitive judgments. Aspects of the results suggest that a basis on which better and poorer rememberers differ is the degree to which certain memory dynamics are understood, such as the fleeting nature of recency effects and the consequences of an initial retrieval. The results have pedagogical as well as theoretical implications, particularly with respect to the education of subjective assessments of ongoing learning.