客体特征统计属性表征机制的特异性

通过两个实验就视觉系统能否像计算平均数那样高效地计算其他统计量的问题进行了探讨。实验—保持—组圆的平均大小不变而改变众数，考察平均数估计是否受众数变化的影响；实验二采用与实验一相同的刺激，直接考察被试估计众数的绩效。结果发现：（1）对平均数的估计不受总体众数变化的影响；（2）对众数的估计往往不如对平均数的估计准确，估计值受平均数变化的影响；（3）在估计众数的任务中，被试成绩受总体极大值的影响。上述结果表明，视觉系统不存在针对众数的自动化加工机制。根据本实验的结果可以进一步推测，视觉系统并非对所有统计量都可做高效加工，而可能存在针对平均数的特异加工机制。

Representation of Statistical Properties: A Specific Mechanism of Visual Perception

Our visual system can extract summary statistics from a group of objects, even when the group is outside the focus of attention. Most previous studies focused on the mechanism of computing the mean size of objects, indicating the visual system is able to exact mean size precisely and efficicnt y. However, it is still unknown whether the visual system can also exact other statistics with the same efficiency as the mean value. In the current research,＇ we put mode size into test, by exploring whether the estimation of mode size could be affected by the change of mean size. Observers were asked to estimate the mean size （Experiment 1 ） and mode size （Experiment 2） of 20 circles presented briefly on the screen.The distribution of size was manipulated to vary the mean size and the mode size. The results show that the mean size estimation was immune to the change in mode size. In contrast, the estimations of mode size were less accurate, and varied with the mean size. Moreover, the mode size was significantly overestimated, especially when the group had more big circles. These results suggest the visual system is not able to extract all summary statistics as efflciently as mean value. The visual system might have a specific mechanism for computing the statistical mean.