Abstract
Eye fixations allow the viewer to visually perceive with high spatial acuity. Here, we tested the role of fixations for scene memory. If fixations drive scene memory, a viewer should repeat a previous fixation pattern when recognizing a scene (Noton & Stark, 1971; Underwood, Foulsham, & Humphrey, 2009). However, salience of low-level features can also account for similarities between two successive fixation patterns within the same images, by attracting the eyes in a stimulus-driven, task-independent manner (Kaspar & König, 2011). Therefore, we tested whether the viewer’s aim to recognize a scene fosters fixations that repeat from learning to recognition. In Experiment 1 we varied the viewing task (recognition vs. free-viewing) between viewers, and compared the extent of repeated fixation patterns between the groups. By showing the same images in both tasks, salience was held constant. We found that the similarity of successive fixation patterns was significantly higher under recognition than under free viewing conditions. Furthermore, viewers fixated significantly more on repeated scene content during recognition but not during free viewing. In Experiment 2 we applied a gaze-contingent presentation of images to test whether participants remembered visual input that they fixated during learning better than salient but non-fixated visual input. To that end, we presented viewers with much smaller cutouts (at screen center) from the previously learned full screen scenes. We found that cutouts featuring previously fixated scene content were recognized better and faster than cutouts featuring non-fixated but highly salient scene content. We conclude that fixations serve visual memory for natural scenes over and above a stimulus-driven influence of visual salience, and discuss how our results relate to memory-based effects, such as priming, reported in the visual search literature.
Meeting abstract presented at VSS 2013