Judgments of learning are affected by the kind of encoding in ways that cannot be attributed to the level of recall.

The authors investigated the theoretical question of whether different kinds of encoding can affect judgments of learning (JOLs) beyond any indirect effects arising from the differences those kinds of encoding produce on the likelihood of recall. They found that JOLs were more accurate after encoding by means of intentional learning than after encoding by means of incidental learning, even when the likelihood of recall did not differ for those kinds of encoding (Experiment 1), and were more accurate when intentional encoding occurred by generating the responses than by reading the responses (Experiment 2). An aggregation effect for JOLs was also discovered: Making JOLs about the likelihood of recall for an aggregate of items yielded less overconfidence (and even underconfidence) in contrast to the typical overconfidence of item-by-item JOLs. The overall pattern of findings suggests that JOLs are theoretically rich and are based on more than whatever underlies the likelihood of recall.