This study investigated the effects of self-generated questions and external questions on thinking processes. Thirty-three college students acted as investigators in a computer simulation of a Congressional investigation into the Pearl Harbor attack. The simulation--known as "The Attack on Pearl Harbor: Cloud of Mystery?"--presented the background of the attack and information based on a 1946 Congressional investigation. At the end of the program, subjects were asked to draw conclusions about who should be blamed for the lack of preparedness on the part of the United States military. Conclusions were scored based on the thinking processes involved, and scores were used as a dependent variable. Subjects received the program under three conditions of questioning: (1) externally-generated high order questions (questions generated by the experimenter that require learners to respond by applying, analyzing, synthesizing, and evaluating the previously presented information); (2) externally-generated low order questions (questions generated by the experimenter that require learners to respond by recalling the information previously presented); and (3) self-generated questions (questions generated by learners in response to the experimenter's prompts). Results indicate that the question conditions were unrelated to the quality of the final explanations. However, notetaking was related to the quality of the final explanations. The non-significant findings for question conditions were interpreted as indicating that questions may elicit performance without the subjects actually interrogating the material. Three data tables are provided. (TJH)