| The Underconfidence with Practice (UWP) effect (Koriat, Sheffer & Ma'ayan, 2002), found in multi-trial learning, is marked by a pattern of underconfidence accompanied by an increase in resolution between the judgments and test on and after the second trial. The experiments presented here tested memory for past test (MPT) heuristic as an explanation of the resolution and calibration effects. According to the MPT heuristic, in the absence of better predictors people may base their judgments of learning (JOLs) on their memory for whether they correctly answered an item on a previous test, and underweight new learning. Five experiments showed evidence that supports use of this heuristic. In Experiment 1 immediate and delayed JOLs were contrasted because while the former show the UWP effect the latter do not. As predicted by use of the MPT heuristic, in the immediate JOL condition an item's Trial 1 test performance better predicted its Trial 2 JOL than did its Trial 2 test performance. Items in the delayed condition showed the reverse pattern. Experiment 2 offered a further test of the MPT heuristic. The number of repetitions (2a) or the study time (2b) on Trial 1 was manipulated, but then the manipulation was reversed on Trial 2, thereby equating performance. Despite the equivalent Trial 2 recall performance, Trial 2 JOLs reflected the manipulated Trial 1 test performance, providing support for the MPT hypothesis. Follow up experiments tested and ruled out several alternative explanations. These experiments support the proposal that the MPT heuristic is involved in second trial immediate JOLs, and that its use may be responsible for the UWP effect. |
