Visual Attention and Preferences of Parents in Selecting Picture Books for Young Children: An Eye-Tracking Study

Abstract

Picture books have become a fundamental medium for emotional expression and cognitive development in early childhood. Parents’ preferences and visual attention during shared reading directly shape children’s exposure to visual narratives. However, few studies have investigated the characteristics of parents’ visual exploration. This proof-of-concept study presents a data-driven intelligent analysis framework that uses mixed-reality eye-tracking, scan-path and heatmap visualization, clustering, and questionnaire-based evaluation to investigate how parents inspect and assess children’s illustrations. Nineteen parents of children aged 1–8 years viewed six groups of composite illustrations through a HoloLens 2 device and subsequently completed a preference questionnaire. For each stimulus, scan paths were grouped using TimeSeriesKMeans with Dynamic Time Warping (DTW), while heatmaps were clustered by K-means using features that describe spatial overlap and attention-density distribution. Based on domain knowledge, candidate values of K from 2 to 4 were considered, and a four-cluster solution was retained for each image in both analyses. Questionnaire results showed that vivid characters and bright, vibrant colors were the most frequently reported positive attributes, whereas unappealing themes, lackluster colors, and abstract characters were the most frequent reasons for rejection. Visual examination of cluster-mean heatmaps indicated concentration around recognizable characters and narrative centers, showing some descriptive correspondence with the questionnaire responses at the aggregate level. However, darker or structurally salient regions without positive evaluations were emphasized by several clusters. This divergence suggests that gaze behavior reflects information acquisition and visual salience rather than preference alone. Temporal and spatial gaze patterns and questionnaire responses are therefore presented as complementary but analytically separate sources of information for exploratory illustration assessment.