Scene gist and local meaning guide attention and build memory representations: Evidence from mouse-clicks

Resumen

Global and local semantic information are thought to influence where attention is allocated in real-world scene images. How the two interact to guide attention and what subsequent effects each has on memory for a viewed scene is unknown. We used the BubbleView paradigm (Kim, Bylinskii, et al., 2017) to monitor movements of attention via participants’ clicks during a scene-viewing task. We manipulated access to scene gist by providing a preview of the BubbleView scene on some trials. ‘Click Maps’ were compiled from participants’ click locations. These were compared to fixation density maps for the same scenes and to meaning maps. Click maps were similar to fixation density maps, suggesting that clicks in the BubbleView paradigm approximate the locus of overt attention. In both a between-subjects (Experiment 1) and a within-subjects comparison (Experiments 2 and 3), access to gist guided attention to locally meaningful scene regions. Local meaning also guided attention, even in the absence of gist. These results support theories describing cognitive guidance of attention in natural scene images. Access to gist and attention to meaning were also important for participants’ memory for the scene images: when participants had access to gist and when they had clicked more meaningful scene regions, they were more likely to remember the image at test. Taken together, our results suggest that global (scene gist) and local scene semantics influence both where we attend in a scene and our subsequent memory for it.

Áreas y palabras clave

Cómo citar (APA 7)

Cronin, D., Choe, K. W., & Henderson, J. M. (2023). Scene gist and local meaning guide attention and build memory representations: Evidence from mouse-clicks. https://doi.org/10.31234/osf.io/dxhz4