Fischhoff (1975, 1977) demonstrated that if people first make a prediction about the outcome of an event and then, at some later time, are given feedback about the actual outcome and are asked what they said originally, they bias their memory for their original judgments in the direction of the actual outcome; that is, they exhibit a phenomenon called hindsight bias or the "knew it all along" effect. For example, Arkes, Wortmann, Saville, and Harkness (1981) provided physicians with a case history followed by laboratory investigations. The physicians were then asked to provide probabilities for four possible diagnoses. The averages given by physicians were 44%,29%, 16%, and 11 % for each of the four. However, a separate group, who were told the diagnosis first (manipulated to equally represent the four possibilities), gave probability estimates of 39%, 35%, 38%, and 31 %. Notably, on the two least likely diagnoses, in particular, the physicians were unable to look at the evidence in a manner unbiased by what they "knew" to be true.