There is ample evidence that emotions affect performance. Positive emotions can improve performance, while negative ones may diminish it. For example, the fears induced by the possibility of failure or of negative evaluations have physiological consequences (shaking, loss of concentration) that may impair performance in sports, on stage or at school. There is also ample evidence that individuals have distorted recollection of past events, and distorted attributions of the causes of successes of failures. Recollection of good events or successes is typically easier than recollection of bad ones or failures. Successes tend to be attributed to intrinsic aptitudes or own effort, while failures are attributed to bad luck. In addition, these attributions are often reversed when judging the performance of others. The objective of this paper is to incorporate the first phenomenon above into an otherwise standard decision theoretic model, and show that in a world where performance depends on emotions, biases in information processing enhance welfare.