Visual attention in interpretive settings is shaped not only by image saliency but also by the context viewers receive before viewing. However, how different context formats relate to attention allocation in curated visual scenes remains under-specified, and findings are often sensitive to the evaluation metric used. We conducted an exploratory free-viewing eye-tracking study in which participants viewed keyframes from museum-oriented scenes after brief pre-viewing context presented in different formats. The dataset comprises gaze recordings from 10 participants across 30 images. Attention patterns were evaluated by comparing observed gaze distributions with a saliency-model baseline and summarizing correspondence through a multi-metric profile. The results indicate systematic block-level differences in gaze-reference correspondence and variability, suggesting that pre-viewing context can be reflected in where attention concentrates, while different metrics highlight different aspects of deviation. These findings offer a compact, metric-aware way to report context-associated attention differences in small-sample studies and inform interpretive design decisions where textual and graphical cues are used to shape viewing.
Li et al. (Mon,) studied this question.