Exploring the Relation between Confidence and Accuracy in Recognition Memory

Recognition memory is examined by exposing a person to a stimulus and later prompting them with the same stimulus to examine their ability to accurately acknowledge that the stimulus was previously encountered (Kahana, 2012). In recognition memory, confidence ratings are taken during the testing phase to assess how confident the participant is that the old-new judgment that they just made is accurate (Busey et al., 2000). Confidence is a metacognitive assessment about the accuracy of perception of decision making based on the amount, speed, and clarity of thoughts that come to mind (Dunlosky and Metcalfe, 2008). The goal of the current study is to better understand how assessing recognition memory using a variety of test procedures influences memory accuracy using the signal detection theory and adding multiple confidence scales that vary in granularity. Based on the previous literature, it is hypothesized that; 1) tasks ordered sequentially will produce greater recognition accuracy (d') than the simultaneous (dual task) condition; 2) confidence scale of 3 points will produce a larger d' than the 7 point scale, and the 7 point scale will produce a larger d' than the 100 point scale; and 3) task mode (ordered vs. sequenced) will interact with confidence scale granularity to predict memory accuracy, such that sequential judgments lessen demands on working memory that come from maintaining an increasing number of decision criteria in comparison to the dual task. Results indicated all hypotheses were not upheld. The findings suggest that taxing working memory may not affect decisional accuracy on a recognition task incorporating confidence judgments.