Cognitive Bases of Hindsight Effects

后见之明偏差的特征是，在一件事情发生后，人们夸大地相信，人们可以事先预测到它。这种“周一早晨的四分卫”被发现损害了包括医学、法律和会计在内的多个领域的判断。例如，了解到正确诊断的医生会大大夸大他们做出诊断的可能性。这种偏见干扰了学习。能够认识到自己可能没有做出正确诊断的医生会更愿意从疑难病例中吸取教训。这项研究解决了后见之明偏差的原因。一个假设的原因是，当一个正确的答案被揭示或发生了一件事时，认知过程的“流畅性”就会增加。例如，如果X发生了，我会花一些时间思考X。当被问到我是否会预测到X、Y或Z会发生时，对X进行的增强处理使其流利性比对Y或Z的流利性更高。为了测试提高的流畅性是否会导致后见之明偏差，我通过一些非正统的方法来控制流利性，然后评估后见之明偏差的程度。例如，在一项研究中，印刷报表的视觉清晰度将受到操纵。这一声明将被如实地贴上“真”或“假”的标签。受试者将被问到，如果他们没有被告知真实情况，他们是否会将这一陈述评为真或假。我预测，当陈述打印得不那么清晰时，它们会产生更少的事后效应，因为不清楚的陈述会降低它们被处理的流利性。另一组实验将测试，与知识较少的人相比，对某个主题了解得更多的人是否更难隔离关于该主题的新知识。例如，教授必须忽略他或她的大部分知识，才能教给正在上课的新手。这类似于事后诸葛亮：在事后诸葛亮的情况下，为了做出准确的评估，一个人必须忽略对结果的了解。在教授的情况下，他或她必须忽略内容知识。我将研究使人们或多或少能够成功地完成这项任务的条件。

Hindsight bias is characterized by an exaggerated belief, following the occurrence of an event, that one could have predicted it beforehand. This 'Monday morning quarterbacking' has been found to compromise judgment in a large variety of domains including medicine, law, and accounting. For example, physicians who learn a correct diagnosis significantly exaggerate the likelihood that they could have arrived at that diagnosis. This bias interferes with learning. A physician who is able to recognize that he or she might not have made a correct diagnosis will be more willing to attend to the lessons to be learned from a difficult case.This research addresses the causes of the hindsight bias. A hypothesized cause is the increased 'fluency' of cognitive processing that occurs when a correct answer is revealed or an event that has occurred. For example, if X occurs, I spend some time thinking about X. When asked whether I would have predicted that X, Y, or Z would occur, the augmented processing done on X makes its fluency higher than that for Y or Z. To test whether heightened fluency causes hindsight bias, I manipulate fluency by some unorthodox means and then assess the magnitude of the hindsight bias. For example, in one study the visual clarity of a printed statement will be manipulated. The statement will be truthfully labeled as either 'true' or 'false.' Subjects will be asked whether they would have rated the statement as true or false if they had not been told the actual truth status. I predict that when statements are printed less clearly they will produce less of a hindsight effect, because the unclear presentation will lower the fluency with which they are processed.Another group of experiments will test whether people more knowledgeable about a topic are less able to quarantine new knowledge about that topic compared to those less knowledgeable. For example, a professor must ignore much of his or her knowledge in order to teach the novices taking a class. This is similar to the hindsight bias: in hindsight-bias situations, a person must ignore knowledge about outcomes in order to make accurate assessments. In the professor's situation, he or she must ignore content knowledge. I will examine the conditions that enable people to perform this task more or less successfully.