Just saw this over at Not Exactly Rocket Science: http://scienceblogs.com/notrocketscience/2010/02/quicker_feedback_for_better_performance.php

Quick summary: They asked a bunch of people to give a 4-minute presentation, had people judging, and told the presenter how long it would be before they heard their assessment. Anticipating quicker feedback resulted in better performance, but predictions of worse performance, and anticipating slower feedback had the reverse effect.