注意分配对大学生数字线估计中左位数效应的影响

The Impact of Assignment on the Left-Digit Effect in College Students’ Estimates of Digital Lines

摘要: 左位数效应(left digit effect)是指多位数中最左侧数位的数值对整体数量估计产生不成比例影响的现象。已有研究在数字线估计任务中反复验证了该效应的稳健性，但对其认知机制的探讨仍相对有限，且鲜有研究从注意分配的角度探索该效应的可干预性。本研究基于符号模型关于序列加工优先性的理论视角，通过操纵数字各位数的呈现时序，考察增加右侧数字注意对减小大学生左位数效应的影响。实验招募90名在校大学生，随机分配至三种实验条件：(1) 无延迟条件(三位数字同时呈现)；(2) #延迟条件(百位数字以“#”占位延迟1 s呈现)；(3) 0延迟条件(百位数字以“0”占位延迟1 s呈现)。延迟条件下，被试需在完整数字呈现前出声报告先呈现的右侧数字，以确保注意被有效引导至右侧位数。因变量为一百边界差异分数(左位数效应大小)，五十边界差异分数作为对照指标。单样本t检验结果表明，各实验条件下的一百边界差异分数均显著大于0 (p < 0.001)，五十边界差异分数均与0无显著差异(p > 0.15)，证实左位数效应在各条件下均稳定存在。单因素方差分析显示，实验条件主效应显著，F(2, 87) = 22.782, p < 0.001, ηp² = 0.409。事后比较表明，无延迟条件下的左位数效应(M = 28.622)显著高于左侧延迟条件(M = 9.249, p < 0.001)和0左侧延迟条件(M = 15.285, p < 0.001)，且左侧延迟条件下的左位数效应显著低于0左侧延迟条件(p = 0.043)。配对样本t检验发现，在延迟组中，左侧延迟条件下的左位数效应显著小于右侧延迟条件(t(29) = −4.882, p < 0.001)；在0延迟组中，左侧延迟条件下的左位数效应亦显著小于右侧延迟条件(t(29) = −2.180, p = 0.037)。上述结果表明，通过操纵呈现时序引导注意至右侧数字可有效削弱左位数效应，且占位符类型和注意方向均对干预效果具有显著调节作用。研究结果为符号模型提供了关键行为证据，提示左位数效应的产生更多与序列加工中的注意分配惯性有关，而非整体认知努力不足。

Abstract: The left digit effect refers to the phenomenon whereby the leftmost digit of a multi-digit number disproportionately influences overall magnitude estimation. Although this effect has been robustly demonstrated in number line estimation tasks, its cognitive mechanisms remain debated, and few studies have explored its modifiability from an attentional perspective. Drawing on the symbol model’s assumption of sequential processing priority, this study manipulated the temporal order of digit presentation to examine whether directing attention to right-side digits attenuates the left digit effect in college students. Ninety undergraduate students were randomly assigned to one of the three conditions: (1) no-delay condition, in which three-digit numbers were presented simultaneously; (2) # delay condition, in which the hundreds digit was initially replaced by “#” and delayed for 1 s; and (3) 0-delay condition, in which the hundreds digit was initially replaced by “0” and delayed for 1 s. In delay conditions, participants were required to orally report the right-side digits before the full number appeared, ensuring that attention was directed to the right. The dependent variable was the hundred-boundary difference score, with fifty-boundary difference scores serving as a control. One-sample t-tests confirmed that hundred-boundary difference scores were significantly greater than zero across all conditions (p < 0.001), whereas fifty-boundary difference scores did not differ from zero (p > 0.15), confirming the robust presence of the left digit effect. A one-way ANOVA revealed a significant main effect of condition, F(2, 87) = 22.782, p < 0.001, ηp² = 0.409. Post-hoc comparisons indicated that the left digit effect in the no-delay condition (M = 28.622) was significantly larger than in the left-delay condition (M = 9.249, p < 0.001) and the 0-left-delay condition (M = 15.285, p < 0.001). Moreover, the effect was significantly smaller in the left-delay condition than in the 0-left-delay condition (p = 0.043). A paired-sample t-test revealed that, in the delay group, the left-digit effect under the left-side delay condition was significantly smaller than that under the right-side delay condition (t(29) = −4.882, p < 0.001); in the zero-delay group, the leftdigit effect under the left-side delay condition was also significantly smaller than that under the right-side delay condition (t(29) = −2.180, p = 0.037). These findings indicate that directing attention to right-side digits via temporal manipulation effectively attenuates the left digit effect, with both placeholder type and attentional direction significantly modulating the magnitude of the effect. The results provide key behavioral evidence supporting the symbol model and suggest that the left digit effect stems more from attentional inertia in sequential processing than from insufficient overall cognitive effort.