摘要：Studies on inference have shown that people use a variety of inference strategies depending on the situation. Despite a great deal of discussion on the use of these strategies at an individual level, very little research has examined how the strategies people use affect group performance. To address this issue, we conducted two computer simulation studies on group decision-making. Our focus was primarily the diversity of strategies used in groups, as previous studies have suggested that diversity plays a critical role in the wisdom of crowds. Therefore, we systematically manipulated the diversity of inference strategies among group members and examined the effect on group performance. In Study 1, we conducted computer simulations using behavioural data from a previous study and found that diversity of strategies could improve group performance. That is, the group whose members used diverse strategies had higher accuracy than groups where all members used an identical strategy. We also investigated how such a phenomenon emerged. In Study 2, we created multiple hypothetical environmental settings and examined the effect. The environmental settings in Study 1 was limited to the ‘kind’ setting, in which correct inferences could be made for most problems by using a certain strategy, and the results of Study 2 showed that the findings of Study 1 could be generalized to other settings. For example, diversity could improve group performance in the ‘wicked’ environment where an inference strategy tends to lead an individual to the wrong answer. We also identified conditions in which the diversity enhanced group performance in each environment. Finally, for Study 1, we conducted additional simulations and discussed the conditions in which diversity would improve group performance more. The contributions to the research on the wisdom of crowds and human inference are discussed.
