Understanding visual-working-memory encoding as a visual search for multiple targets

点击翻译按钮获取中文摘要

Experience with the environment influences conscious perception by biasing the selection of information. Most conscious processing of visual information goes via a system called visual working memory (VWM). VWM is severely limited in its capacity so that only a tiny fraction of the information available in the environment is encoded and represented at a time. Furthermore, even among those objects that made it into VWM, representational quality varies considerably. The reasons for this inter-item variability remain largely unknown. I suggest that a huge share of this variability emerges during encoding: instead of passively absorbing all relevant information until VWM capacity is exhausted, VWM encoding might be characterized by an active interaction between the visual scene (e.g., different saliencies of the objects) and the observer (e.g., experience with the environment: predictive coding). By examining this idea, the project aims to close a crucial gap in the VWM literature and to extend the scope of the research unit to the next step in the processing chain: active selection of visual information for further processing (maintenance).The project relates two central attentional mechanisms to VWM encoding: priority maps and focal spatial attention. A priority map is a spatial representation of the visual scene that abstracts from specific features of objects and instead provides a single priority value at each location, reflecting a combination of bottom-up saliency, top-down goals, and previous experiences (task history). The main hypothesis tested in WP1 is that the priority map determines the quality of encoding into VWM. This will be examined via (experimental) variations in the two contributors to the priority map that were so far largely neglected in VWM research, namely the objects’ saliencies (WP1a) and task history (WP1b). WP2 will examine whether focal spatial attention contributes to VWM encoding by upgrading one object into a special state. This special state has been the subject of vigorous debates and extensive empirical investigation, but it is as yet unknown what brings an object into that state if it is not explicitly cued and in how far being in the special state is qualitatively different from having the highest priority.Using elements from various validated models, a new computational model of VWM encoding will be developed to explain the expected complex data patterns in considerable detail and to gain a principled understanding of the cognitive mechanisms involved. Furthermore, electrophysiological markers of attention allocations that were previously validated for visual search will be used to understand the attentional dynamics that influence VWM encoding and will also be integrated into the model. In sum, the present project will examine whether and how various mechanisms of active perception affect object processing beyond attentional selection.