Abstract
To interpret colormaps, people determine how dimensions of color map onto quantities. Colormaps are easier to interpret when encoded mappings match people’s predicted mappings, but to harness this principle it is necessary to understand predicted mappings. Previous work demonstrated evidence for dark-is-more and opaque-is-more biases for grid colormaps—faster response times for interpreting colormaps when darker, more opaque colors mapped to larger quantities (Schloss, et al, 2019). However, it is unknown whether such biases persist when spatial configurations cue which regions represent larger quantities (e.g., hotspots in concentric configurations) (Schott, 2010). We investigated how the dark-is-more bias is influenced by spatial organization by comparing people’s interpretations of grid and concentric colormaps. Colormaps displayed fictitious data on animal sightings across geographical regions and participants reported whether there were more sightings on the left/right. One side was always darker (left/right balanced), and the legend either encoded dark-more or light-more mapping. The design included 2 spatial configurations (grid/concentric) × 2 color scales (hot/viridis) × 2 hotspot lightnesses (dark/light) × 2 encoded mappings (dark-more/light-more) × 2 legend text positions (greater-high/greater-low) × 20 repetitions (different underlying datasets). Trials were blocked by spatial configuration (order counterbalanced). Overall, response times were faster for dark-more encodings, indicating a dark-is-more bias (p< .01), but this depended on spatial configuration and block order. Comparing “pure” grid and concentric conditions (each in block 1), a mapping × spatial configuration interaction (p< .05) indicated that the dark-is-more bias for grids (p< .001) was reduced for concentric configurations. However, exposure to spatial configuration in block 1 influenced responses in block 2. The dark-is-more bias for grids in the grid-first condition transferred to concentric configurations (p< .01), and the reduced dark-is-more bias for concentric configurations in the concentric-first condition transferred to grids. Therefore, spatial organization does affect interpretations of colormaps, as do contextual colormap configurations.
Acknowledgement: Wisconsin Alumni Research Foundation (WARF)