In hindsight bias, upon learning an outcome, one is overly confident that one would have “known it all along.” Several researchers have been able to neutralize hindsight bias by prompting participants to consider alternative outcomes, but can we learn to avoid bias for novel outcomes, without prompting? Foresight participants read brief summaries of five psychology studies, and learned the mean performance of one group in each study. They estimated the other group’s performance—reflecting their sense of the effect size—stated possible causes, and then learned the other group’s mean performance. Hindsight participants learned both groups’ mean performance at the outset, then indicated what they would have estimated. We asked whether (1) participants would show superior estimation and/or consideration of alternative causes for novel stimuli one week later, and (2) whether Foresight participants would benefit more given the feedback they received on the accuracy of their estimates.