Feedback effects in a metric multiple-cue probability learning task

Abstract

Three forms of augmented feedback were compared with simple outcome feedback in a computer-run, metric, multiple-cue probability learning task. These were: (a) lens model feedback, (b) provision of the predicted criterion value, and (c) provision of categorical information about the criterion. There was significant improvement in performance across six blocks of 25 trials, but no differences among the feedback treatments. In a second study, two additional groups were run without outcome feedback. These groups received either lens model feedback or predicted criterion feedback. The major result was a sharp increase in consistency, which mediated an increase in achievement, when outcome feedback was not provided.