Visual narratives like comics and films often shift between showing full scenes and close, zoomed-in viewpoints. These zooms are similar to the “spotlight of attention” cast across a visual scene in perception. We here measured ERPs to visual narratives (comic strips) that used zoomed-in and full-scene panels either throughout the whole sequence context or at specific critical panels. Zoomed-in panels were automatically generated on the basis of fixations from prior participants’ eye movements to the crucial content of panels (Foulsham & Cohn, 2020). We found that these fixation panels evoked a smaller N300 than full-scenes, indicative of reduced cost for object identification, but that they also evoked a slightly larger amplitude N400 response, suggesting a greater cost for accessing semantic memory with constrained content. Panels in sequences where fixation panels persisted across all positions of the sequence also evoked larger posterior P600s, implying that constrained views required more updating or revision processes throughout the sequence. Altogether, these findings suggest that constraining a visual scene to its crucial parts triggers various processes related not only to the density of its information but also to its integration into a sequential context.

诸如漫画和电影之类的视觉叙事通常会在展现完整场景和放大近距离视点之间转移。这些缩放类似于感知中跨视觉场景的“关注焦点”。在这里，我们将ERP评估为视觉叙事（漫画带），在整个序列上下文中或在特定的关键面板上使用放大和全屏面板。放大的面板是根据先前参与者的眼动到面板关键内容的注视自动生成的（Foulsham＆Cohn，2020）。我们发现，这些固定面板引起的N300小于完整场景，表明对象识别的成本降低，但是它们引起的振幅N400响应也稍大，这表明访问内容受限的语义记忆的成本更高。序列中的固定面板跨序列的所有位置持续存在的面板也引起较大的后部P600，这意味着受约束的视图在整个序列中需要更多的更新或修订过程。总而言之，这些发现表明，将视觉场景约束到其关键部分会触发各种过程，这些过程不仅涉及其信息的密度，而且还涉及其整合到顺序环境中。