Studying attention to IPCC climate change maps with mobile eye-tracking

viewing condition compared to the paired: MSingle = 154.87 (±7.07), MPaired = 216.86 (±20.27). Additionally, the average fixation duration was only slightly longer in the single viewing condi- tion: MSingle = 380.51 ms (±11.41), MPaired = 359.95 ms (±18.68). The proxy scanpath length, measured as the cumulative sum of Euclidean distances between successive fixations based on reference images (685 × 1000 pixels for the Baptism of Christ (Painting #1) and 983 × 1000 pix- els for The Swan (Painting #2)), was significantly shorter in the single viewing condition: MSin- gle = 24196.75 px (±1205.88), MPaired = 37468.70 px (±3332.50), refer to Fig 4 for an overview as box plots and S6 Table for aggregate metrics, divided by viewing two conditions. Despite the relatively low sample size and the exploratory nature of the research, a nonpara- metric ANOVA (Kruskal-Wallis test) was employed for statistical analysis. While average fixa- tion duration showed no significant differences, total fixation duration, fixation count, and scanpath length exhibited significant differences across conditions (see S7 Table for a detailed breakdown of χ2, df, p, ε2. Similar trends were observed when the data were further disaggre- gated by the two paintings (see S7 Fig and S8 Table). Although not subjected to statistical comparison, it may be useful to highlight observed trends between map and painting viewing. On average, participants spent about twice as much time viewing paintings compared to maps, as indexed by dwell time. This discrepancy may be

attributed to several factors: Participants might have found artworks more engaging than cli- mate crisis maps, or the smaller number of paintings (two) compared to maps (ten) could have allowed for longer individual viewing times per painting. Additionally, the maps might not have contained as much visual information or complexity, which might have required less time to view. Another notable difference was also observed in average fixation duration, which was highly shorter for painting viewing compared to map viewing. Longer average fixation durations sometimes suggest a higher cognitive load, while shorter fixations might indicate that the task of viewing paintings was less cognitively demanding, or that the information was Fig 4. Box plots of gaze metrics for paintings. Box plots depict five gaze metrics for two paintings (single-viewing in light grey and paired-viewing in dark grey): (a) total fixation duration (s), (b) normalised fixation duration (%), (c) fixation count, (d) average fixation duration (ms), and (e) proxy scanpath length (pixels). On average, single viewing showed slightly shorter dwell times, lower fixation counts, slightly longer average fixation durations, and significantly shorter scanpath lengths. Each plot shows the range (excluding outliers), interquartile range, median, and mean (triangle overlay). https://doi.org/10.1371/journal.pone.0316909.g004 PLOS ONE Studying attention to IPCC climate change maps with mobile eye-tracking PLOS ONE | https://doi.org/10.1371/journal.pone.0316909 January 10, 2025 10 / 17