Pervasive congruency effects characterize approximate number discrimination tasks. Performance is better on congruent (the more numerous stimulus consists of objects of larger size that occupy a larger area) than on incongruent (where the opposite holds) items. The congruency effects typically occur when controlling for nonnumeric variables such as cumulative area. Furthermore, only performance on incongruent stimuli seems to predict math abilities. Here, we present evidence for an attentional-bias induced by stimulus control (ABC) where preattentive features such as item size reflexively influence decisions, which can explain these congruency effects. In three experiments, we tested predictions derived from the ABC. In Experiment 1, as predicted, we found that manipulation of size introduced congruency effects and eliminated the correlation with math ability for congruent items. However, performance on incongruent items and neutral, nonmanipulated items were still predictive of math ability. A negative correlation between performance on congruent and incongruent items even indicated that they measure different underlying constructs. Experiment 2 demonstrated, in line with the ABC account, that increasing presentation time reduced congruency effects. By directly measuring overt attention using eye-tracking, Experiment 3 revealed that people direct their first gaze toward the array with items of larger individual size, biasing them towards these arrays. The ABC explains why the relation between performance on approximate number discrimination tasks and math achievement has been fragile and suggests that stimulus control manipulations have contaminated the results. We discuss the importance of using stimuli that are representative of the environment.

普遍的一致性效应表征了近似数判别任务。在一致（更多的刺激由占据更大面积的更大尺寸的物体组成）上的表现比在不一致（相反的情况下）项目上的表现更好。在控制非数值变量（例如累积面积）时，通常会出现一致性效应。此外，似乎只有在不一致刺激下的表现才能预测数学能力。在这里，我们提供了由刺激控制 (ABC) 引起的注意力偏差的证据，其中项目大小等前注意力特征会反射性地影响决策，这可以解释这些一致性效应。在三个实验中，我们测试了来自 ABC 的预测。在实验 1 中，正如预测的那样，我们发现对大小的操纵引入了一致性效应并消除了与一致性项目的数学能力的相关性。然而，在不一致的项目和中性的、未处理的项目上的表现仍然可以预测数学能力。一致和不一致项目的表现之间的负相关甚至表明它们衡量不同的基础结构。实验 2 表明，与 ABC 帐户一致，增加演示时间会降低一致性效应。通过使用眼动追踪直接测量明显的注意力，实验 3 显示人们将他们的第一个视线定向到具有较大个体尺寸的项目的阵列，使他们偏向这些阵列。ABC 解释了为什么近似数辨别任务的表现与数学成绩之间的关系很脆弱，并表明刺激控制操作已经污染了结果。我们讨论了使用代表环境的刺激的重要性。