Internal mapping and its impact on measures of absolute and relative metacognitive accuracy

Research in decision making and metacognition has long investigated the calibration of subjective probabilities. To assess calibration, mean ratings on a percentage scale (e.g., subjective likelihood of recalling an item) are typically compared directly to performance percentages (e.g., actual likelihood of recall). Means that are similar versus discrepant are believed to indicate good versus poor calibration, respectively. This chapter argues that this process is incomplete: it examines only the mapping between the overt scale values and objective performance (mapping 2), while ignoring the process by which the overt scale values are first assigned to different levels of subjective evidence (mapping 1). The chapter demonstrates how ignoring mapping 1 can lead to conclusions about calibration that are misleading. It proposes a signal detection framework that not only provides a powerful method for analyzing calibration data, but also offers a variety of measures of relative metacognitive accuracy (resolution