Abstract: The present research explored predictive accuracy with a new FOK criterion test─over relearning. In the first experiment, the 18 subjects, 10 men and 8 women, were divided into two groups. The subjects were asked to learn and recall the targets of word digit pairs and then make FOK predictions for all non recalled answers. In the criterion test phase, one group of subjects underwent recognition, while the other group underwent relearning. The numbers of the relearning were recorded. In the second experiments, there were 12 subjects. The materials were a list of 40 word digit pairs from the first experiment. The 40 word pairs were divided into two equal difficulty groups. The experiment procedure was the same as the first experiment except for the criterion test phase. In the criterion test phase, half of the subjects first recognized and then relearned. The other half of the subjects first relearned and then recognized. There was a negative correlation between the FOK rate and the number of the relearning. Both experiments demonstrated that over relearning can be employed successfully as a criterion test. Both experiments showed that not only the feeling of knowing was greater for commission error items than for omission errors, but also the criterion test performance which contained recognition and over relearning was better for commission error items than for omission error items.

Key words: over relearning, commission error, omission error, criterion test