Remember-No experiment

Abstract

Remembering is an important but potentially unreliable cue influencing both confidence and accuracy, however, high confidence remembering is believed to be reliably associated with high recognition accuracy. We assessed these beliefs using memory for colour photographs of natural scenes in two separate studies, a 2-alternative forced-choice recognition memory task based on the picture similarity paradigm, and a source memory task with continuous measures of retrieval. We found that confidence in recognition accuracy was influenced more by the experience of remembering than by its accuracy, remembering moderating the association between accuracy and confidence. Likewise, judgments of remembered vividness influenced confidence more than source memory accuracy, memory quality being strongly related to confidence in both studies, with calibration plots confirming that high confidence was associated with high accuracy. In the picture similarity task, high accuracy was associated with remembering with high confidence only when either perceptually or mnemonically similar test-pairs were presented. Notably, when mnemonically similar test-pairs were presented, an association was also found between high accuracy and high confidence in the absence of remembering [speculated due to recollecting (misremembering) a similar picture stored in memory]. Our results demonstrate that remembering with high confidence is not dependably associated with high recognition accuracy. Critically, in the absence of remembering, there was no consistent relationship between accuracy and confidence. Our findings have implications for accounts of vividness, confidence, episodic memory, and eyewitness testimony and offer a novel explanation for the dissociation between accuracy and confidence in the picture similarity task. High confidence remembering may not in all cases denote high recognition accuracy. Highly vivid memories, confidently recollected, may not always denote the factual accuracy of memory.