Jabar Syaheed B, Anderson Britt
Department of Psychology, University of Waterloo.
J Exp Psychol Hum Percept Perform. 2015 Dec;41(6):1666-79. doi: 10.1037/xhp0000121. Epub 2015 Aug 17.
Probability is known to affect perceptual estimations, but an understanding of mechanisms is lacking. Moving beyond binary classification tasks, we had naive participants report the orientation of briefly viewed gratings where we systematically manipulated contingent probability. Participants rapidly developed faster and more precise estimations for high-probability tilts. The shapes of their error distributions, as indexed by a kurtosis measure, also showed a distortion from Gaussian. This kurtosis metric was robust, capturing probability effects that were graded, contextual, and varying as a function of stimulus orientation. Our data can be understood as a probability-induced reduction in the variability or "shape" of estimation errors, as would be expected if probability affects the perceptual representations. As probability manipulations are an implicit component of many endogenous cuing paradigms, changes at the perceptual level could account for changes in performance that might have traditionally been ascribed to "attention."
已知概率会影响感知估计，但尚缺乏对其机制的理解。超越二元分类任务，我们让天真的参与者报告短暂呈现的光栅的方向，在此过程中我们系统地操纵了偶然概率。参与者对高概率倾斜方向迅速形成了更快且更精确的估计。用峰度测量指标衡量，他们误差分布的形状也显示出偏离高斯分布。这种峰度指标很稳健，捕捉到了分级、情境化且随刺激方向变化的概率效应。我们的数据可以理解为概率导致估计误差的变异性或“形状”降低，如果概率影响感知表征，这是可以预期的。由于概率操纵是许多内源性线索范式的一个隐含组成部分，感知层面的变化可以解释那些传统上可能归因于“注意力”的表现变化。
