Confidence enhanced performance : does it exist and if so how does it work?

Abstract

We empirically test and investigate the psychological mechanism behind the theory of Confidence Enhanced Performance (Compte and Postlewaite, 2004). This theory suggests that if confidence enhances performance, then people will benefit from having the commonly observed tendency to discount past failures and be overconfident. We test this hypothesis using three tasks which require different cognitive abilities. Our findings are partially supportive of the theory. However, we cannot conclude that any of our confidence variables significantly and positively affect performance. Only in one task (the Picture Recall task) we find that enhanced mood level positively affects performance. The view that people are generally overconfident when performing tasks with imperfect feedback is not upheld. Furthermore, there seem to be strong task differences in both levels of confidence and the effects of psychological variables on performance.