Study on mechanisms of integration of featural and spatiotemporal information in visual cognition of dynamic multidimensional situations

(1)动态多维情境下目标信息的绑定：设计了多目标永久跟踪（Multiple object Permanence Tracking， MOPT）范式，研究了视觉工作记忆中特征信息与颜色、形状等时空信息的整合。虽然有人建议我们可以保存四到五个对象的信息，但我们已经证明，在动态和多维的情况下，我们最多可以保存两个对象。我们视觉工作记忆的功能单元并不是综合的物体表征，特征和位置的综合本身是物体运动和需要记忆的物体数量的函数。(2)注意焦点运动的神经网络模型：建立了视觉搜索任务中注意运动的脉冲神经网络模型。通过在视觉搜索模型中经常使用的显著性图模型中引入随机机制，该模型可以很自然地解释更多的搜索不对称现象。此外，通过心理物理实验验证了该模型对单例搜索任务中搜索不对称性的预测。目前，我们正在扩展该模型，以解释第(1)节中描述的视觉工作记忆实验的发现。(3)视觉表面在视觉工作记忆中的作用：我们使用带有主观表面图形的变化检测任务进行了一系列关于物体表面表征在视觉工作记忆中的作用的实验。在对空白之间形状变化的检测中，当一个变化的物体包含主观表面的形成和破坏时，即使人物的物理变化量相同，表现也更准确。这种优势发生在主观表面的层面，而不是主观边缘的层面。此外，我们使用一种叫做斜坡保持分析的技术来估计记忆容量和处理速度的极限，我们发现主观表面的影响反映了物体表征形成过程中处理速度的差异。少

( 1 ) Binding of object information in dynamic multidimensional situations: I devised a paradigm called Multiple Object Permanence Tracking (MOPT), and investigated the integration of featural and spatiotemporal information such as color and shape in visual working memory. Although it has been suggested that we can hold information of four to five objects, we have shown that in dynamic and multidimensional situations we could hold at most two objects. The functional unit of our visual working memory is not integrated object representations, and the integration of feature and location itself is a function of object motion and the number of object to be remembered.( 2 ) Neural network models for movement of attentional focus: We have built a pulsed neural network models for attentional movements in visual search tasks. By the introduction of stochastic mechanism in a saliency map model that is frequently used in the models for visual search, the model could account for the phenomenon of … More search asymmetry quite naturally. Furthermore, this model's novel prediction, search asymmetry in singleton search tasks, was confirmed by psychophysical experiments. Currently, we are extending the model to account for the findings of the Experiments on visual working memory described in section (1).( 3 ) Roles of visual surfaces in visual working memory: We conducted a series of experiments on the role of surface representation of objects in visual working memory using a change detection task with subjective surface figures. In a detection of shape change between blanks, performance was more accurate when a changing object contains the formation and destruction of subjective surfaces, even if the amount of physical change of figures was identical. This advantage occurred at the level of subjective surfaces, not at the level of subjective edges. Moreover, using a technique to estimate the limits of memory capacity and processing speed called slope-hold analysis, we showed that the effect of subjective surfaces reflects differences in processing speed in formation of object representation. Less