进化频率表征对统计推理(非)灵活性的影响:认知风格和简要提示不影响贝叶斯推理(英文)

The (In)flexibility of Evolved Frequency Representations for Statistical Reasoning: Cognitive Styles and Brief Prompts Do Not Influence Bayesian Inference

当形式操纵有助于贝叶斯推理时会有怎样的加工发生呢?一种观点认为自然取样的频率可以激发在其操作中具有相对特异性的特权表征系统。而与之相对的一种观点则认为,自然取样频率只是引发具有嵌套关系的更为普遍的加工的一种方式。比较两种观点,后者预示着只需要使用相当简要和直接的干预(如简单的指示)就能够促进推理的改善,而前者则意味着更为广泛的干预和/或更有洞见的理解才能改善推理。本研究表明,无论是短暂立即的干预,还是预存的表征偏向,抑或是表征的灵活性都不能促进被试的表现。另一方面,也有证据显示,频率论者的问题解释可以改善统计推理表现,而且有时还会增加其反应的信心。这些结果支持了特权表征系统观。

What happens when format manipulations improve Bayesian reasoning？ One view is that naturally sampled frequencies help induce a privileged representational system that is relatively specific in its operation. A contrasting view is that naturally sampled frequencies are but one way to induce a more general process of appreciating nested set relationships. This later view implies that fairly brief and immediate interventions （e.g., simple directives） should produce improvement, whereas the former view implies that more extensive interventions and/or more insightful understanding are necessary for improvement. The present research indicates that neither brief and immediate interventions nor pre-existing representational biases or representational flexibility facilitate performance. Some evidence emerged, on the other hand, that frequentist problem interpretation can improve statistical reasoning performance and increase confidence in responses at times. These results support the privileged representational system view.