Cui Lucy, Lo Stephanie, Liu Zili
Department of Psychology, University of California, Los Angeles, CA 90095, USA.
Vision (Basel). 2023 Mar 2;7(1):17. doi: 10.3390/vision7010017.
Decisions are often made under uncertainty. The most that one can do is use prior knowledge (e.g., base rates, prior probabilities, etc.) and make the most probable choice given the information we have. Unfortunately, most people struggle with Bayesian reasoning. Poor performance within Bayesian reasoning problems has led researchers to investigate ways to improve Bayesian reasoning. Many have found success in using natural frequencies instead of probabilities to frame problems. Beyond the quantitative format, there is growing literature on the use of visualizations or visual representations to improve Bayesian reasoning, which will be the focus of this review. In this review, we discuss studies that have found visualizations to be effective for improving Bayesian reasoning in a lab or classroom setting and discuss the considerations for using visualizations, paying special attention to individual differences. In addition, we will review the factors that influence Bayesian reasoning, such as natural frequencies vs. probabilities, problem format, individual differences, and interactivity. We also provide general and specific suggestions for future research.
决策往往是在不确定的情况下做出的。人们最多只能利用先验知识(例如,基础比率、先验概率等),并根据我们所拥有的信息做出最有可能的选择。不幸的是,大多数人在贝叶斯推理方面存在困难。贝叶斯推理问题中的糟糕表现促使研究人员去探索改进贝叶斯推理的方法。许多人发现在用自然频率而非概率来构建问题时取得了成功。除了定量形式,关于使用可视化或视觉表征来改进贝叶斯推理的文献越来越多,这将是本综述的重点。在本综述中,我们讨论了那些发现在实验室或课堂环境中可视化对改进贝叶斯推理有效的研究,并讨论了使用可视化时的注意事项,特别关注个体差异。此外,我们将回顾影响贝叶斯推理的因素,如自然频率与概率、问题形式、个体差异和交互性。我们还为未来的研究提供了一般性和具体性的建议。