How does the prevalence of a target influence how it is perceived and categorized? A substantial body of work, mostly in visual search, shows that a higher proportion of targets are missed when prevalence is low. This classic low prevalence effect (LPE) involves a shift to a more conservative decision criterion that makes it less likely that observers will call an ambiguous item a target. In contrast, Levari et al. (Science, 360[6396], 1465–1467, 2018) recently reported the opposite effect in a simple categorization task. In their hands, at low prevalence, observers adopted a more liberal criterion, making observers more likely to label ambiguous dots on a blue–purple continuum “blue.” They called this “prevalence-induced concept change” (PICC). Here, we report that the presence or absence of feedback is critical. With feedback, observers become more conservative at low prevalence, as in the LPE. Without feedback, they become more liberal, identifying a wider range of stimuli as targets, as in Levari’s PICC studies. Stimuli from a shape continuum ranging from rounded (“Bouba”) to bumpy (“Kiki”) shapes produced similar results. Other variables: response type (2AFC vs. go/no-go), color (blue–purple vs. red–green), and stimuli type (solid color vs. texture) did not influence the criterion shifts. Understanding these effects of prevalence and ways they can be controlled illuminates the context-specific nature of perceptual decisions and may be useful in socially important, low prevalence tasks like cancer screening, airport security, and disease diagnosis in pathology.

目标的普遍性如何影响它的感知和分类方式？大量工作（主要是视觉搜索）表明，当流行率较低时，目标丢失的比例较高。这种经典的低流行效应 (LPE) 涉及向更保守的决策标准的转变，这使得观察者不太可能将不明确的项目称为目标。相比之下，Levari 等人。（科学，360[6396], 1465–1467, 2018) 最近在一个简单的分类任务中报告了相反的效果。在他们手中，在低流行率下，观察者采用了更自由的标准，使观察者更有可能将蓝紫色连续体上的模糊点标记为“蓝色”。他们称此为“流行引起的概念改变”（PICC）。在这里，我们报告反馈的存在与否至关重要。有了反馈，观察者在低流行率时变得更加保守，就像在 LPE 中一样。没有反馈，他们变得更加自由，将更广泛的刺激确定为目标，就像 Levari 的 PICC 研究一样。来自圆形（“Bouba”）到凹凸不平（“Kiki”）形状的连续形状的刺激产生了类似的结果。其他变量：反应类型（2AFC vs. go/no-go）、颜色（蓝紫色 vs. 红绿色）和刺激类型（纯色 vs 黑色）。纹理）不影响标准变化。了解流行率的这些影响及其控制方式可以阐明感知决策的特定背景性质，并且可能有助于具有社会重要性的低流行率任务，如癌症筛查、机场安全和病理学疾病诊断。