Recalibration effects in judgments of learning: A signal detection analysis

In this study we investigated the influence of list composition on judgments of learning (JOLs). To this end, we compared JOLs assigned in a multi-cycle procedure to a set of moderately difficult word pairs. Experiment 1 revealed that when difficult new pairs were added to the study list, the mean of JOLs assigned to the moderate pairs increased as compared to the baseline. In Experiment 2, we reversed this pattern by including easy new pairs in the study list. By analyzing metacognitive ROCs (MROCs), we demonstrate that these results were caused by criterion shifts, by which participants adjusted the level of evidence needed to assign particular JOL ratings. Changes in the study list composition led to a recalibration of the JOL scale - i.e. resetting of the criteria - in order to accommodate the addition of new items. We discuss the usefulness of MROCs for detecting criterion shifts in rating tasks.