Garcia-Retamero Rocio, Takezawa Masanori, Gigerenzer Gerd
Max Planck Institute for Human Development, Berlin, Germany.
Exp Psychol. 2009;56(5):307-20. doi: 10.1027/1618-3169.56.5.307.
Inferences are often based on uncertain cues, and the accuracy of such inferences depends on the order in which the cues are searched. Previous research has shown that people and computers progress only slowly in individual learning of cue orderings through feedback. A clue to how people (as opposed to computers) solve this problem is social learning: By exchanging information with others, people can learn which cues are relevant and the order in which they should be considered. By means of simulation, we demonstrate that imitate-the-best and imitate-the-majority speed up individual learning, whereas a third social rule, the Borda rule, does not. Imitate-the-best also leads to a steep increase in learning after a single social exchange, to cue orders that are more accurate than ecological validity, and to faster learning than when individuals gain the learning experience of all other group members but learn without social exchange. In two experiments, we find that people speed up cue learning in a similar way when provided with social information, both when they obtain the information from the experimenter or in free discussions with others.
推理往往基于不确定的线索,而此类推理的准确性取决于搜索线索的顺序。先前的研究表明,人和计算机通过反馈在个体学习线索排序方面进展缓慢。关于人类(与计算机相对)如何解决这一问题的一个线索是社会学习:通过与他人交流信息,人们可以了解哪些线索是相关的以及应该按照何种顺序来考虑这些线索。通过模拟,我们证明“模仿最佳”和“模仿多数”能够加速个体学习,而第三种社会规则——博尔达规则则不然。“模仿最佳”还会在单次社会交流后使学习大幅加速,得到比生态效度更准确的线索顺序,并且比个体获得所有其他群体成员的学习经验但在无社会交流情况下学习时的速度更快。在两项实验中,我们发现,当人们从实验者那里获取信息或在与他人的自由讨论中获得社会信息时,他们会以类似的方式加速线索学习。