视觉信息分布调节信息整合策略：来自眼动的证据

以空白单元格定位为实验任务，操纵序列点阵的分布，考察视觉短时记忆与视知觉的信息整合机制。实验一发现任务正确率与点阵1的圆点数呈近似U形关系。实验二发现在拐点前后被试的眼动模式从偏向于注视圆点位置转为偏向于注视空格位置。因此，视觉信息的分布模式调节整合策略，当处于视觉短时记忆中的圆点数在其容量之内时，采用图像—知觉整合策略，反之，则采用转换—比较策略。

Information Distribution Modulates Integration Strategies:Evidence from Eye Movements

In the present study, empty cell localization task was used to explore how information is integrated between visual short-term memory (VSTM) and visual perception. To explain the participants performances of the task, two hypotheses were proposed: image-percept hypothesis and convert-and-compare hypothesis. Image-percept hypothesis suggested that participants firstly encoded the locations of dots in array 1 and its mental image was formed to integrate with array 2. On the contrary, convert-and-compare hypothesis asserted that participants firstly encoded the location of empty cells in array 1 and its negative image was formed to compare with array 2.Analyzing the existing evidence for or against the two hypotheses, we found that the difference in dot distribution might contribute to different integrating strategies. To test this hypothesis, two experiments were conducted. In experiment 1, the dot distribution of sequential arrays (4–11, 5–10, 6–9, 7–8 and 8–7) and interstimulus interval (ISI, 500 ms, 1500 ms and 2500 ms) were manipulated. The results showed that the accuracy was a U-shaped function of array 1 dot number and at the condition of 6–9 distribution the accuracy reached a lowest value. We argued the findings suggested that integration strategies might change from “memorizing dots” to “memorizing empty cells” of array 1.In experiment 2, participants’ eye movements were monitored when performing empty cell localization task. Dot distribution varied in 3 levels (5–10, 6–9, 7–8) and ISI varied in 2 levels (500ms and 1500ms). The accuracy again showed a U-shaped function of array 1 dot number. In 1500ms ISI condition, there were significant different eye movement patterns between the 3 levels of dot distribution. Specifically, in 5–10 condition, participants showed a bias to look at dot locations during ISI period; in 6–9 condition, participants looked at locations of dots or empty cells on a chance level; while in 7–8 condition, a bias to looking at empty cell location was showed. Thus we found clear eye-movement evidence indicating a change of integration strategy from “memorizing dots” to “memorizing empty cells” in experiment 2. In sum, the present study suggested that human participants are flexible in integrating sequentially presented information because a change of integration strategy was revealed by our data. We argued that the limit of VSTM capacity played an important role in this adaptive characteristic. It seemed that “memorizing dots” strategy was adopted when dot numbers were within the capacity of VSTM while “memorizing empty cells” strategy was adopted when dot numbers were out of VSTM capacity.