Search

In this study we will investigate whether the opaque-is-more bias can be activated for colormap data visualizations when their color scale does not appear to vary in opacity globally. Previous work demonstrated that when interpreting colormap data visualizations, multiple biases influence their inferred mappings between perceptual features and quantities. For example, when colormaps appear to vary in lightness, the dark-is-more bias leads to the inference that darker regions map to larger quantities. And, when colormaps appear to vary in opacity, the opaque-is-more bias leads to the inference that regions appearing more opaque map to larger quantities. When both biases are activated, they work together when colormaps are presented on light backgrounds, but conflict when colormaps are presented on dark backgrounds. Under such conflicts, the opaque-is-more bias can reduce or even override the dark-is-more bias. Previously, activation of the opaque-is-more bias was estimated using an opacity variation index, which was computed over the coordinates of the entire color scale used to generate the colormap (Schloss et al., 2019). Color scales had a perfect opacity variation index if they followed a perfect linear interpolation between the highest contrast color and the background color (e.g., shades of gray on a black or white background). Apparent opacity variation (quantified by the index) decreased as the color scale deviated from this linear interpolation. We refer to the opacity variation index as “global” because it gives equal weight to all of the colors in the color scale. However, a color scale can have weak global opacity variation, while appearing to vary in opacity in “local” segments within the color scale. We propose that in such cases, the opaque-is-more bias can be activated if the underlying dataset used to construct the colormap predominantly maps to colors in the range of strong local opacity variation. Activation of the opaque-is-more bias may be especially strong if regions in the colormap data visualization match the background color, causing regions of the map to appear transparent, or like a perceptual hole (Bartel et al., 2020). In this study, we will test whether colormaps with weak global opacity variation but strong local opacity variation can activate the opaque-is-more bias.