Unifying Effects of Direct and Relational Associations for Visual Communication

Color Semantics in Human Cognition

“…3). For visualizations with > 2 colors and concepts (Schloss et al, 2018) or involving relational rather than discrete concepts (Schoenlein et al, 2023), merit can be more complex. 3 When participants are asked which color represents a target concept (e.g., trash), the concepts and colors in the context of the encoding system are activated within the network.…”

“…These approaches will become more scalable as methods improve for automatically estimating color-concept associations without extensive human judgments (Lin et al, 2013;Rathore et al, 2020). Although color-concept associations are dynamic, updating with experience (Schoenlein et al, 2023), evidence suggests they are sufficiently stable at the group level to predict group-level color preferences (Palmer & Schloss, 2010) and color meaning in information visualizations (Mukherjee et al, 2022;Schloss et al, 2018Schloss et al, , 2021Schoenlein et al, 2023), which is key for designing visualizations for public audiences. The results of these studies can extend to incorporating color semantics into recommender tools for effective visualization design (Gramazio et al, 2017;Smart et al, 2020) and have already contributed to understanding color in marketing (Spence & Van Doorn, 2022).…”

“…3). For visualizations with > 2 colors and concepts (Schloss et al, 2018) or involving relational rather than discrete concepts (Schoenlein et al, 2023), merit can be more complex. 3…”

“… 3. For colormap data visualizations where variations in data magnitude are represented by gradations of color, merit is informed by “direct” color-concept associations between individual colors and concepts represented in the map (e.g., data representing more sunshine are associated with lighter yellowish colors) and “relational” associations (e.g., data representing larger magnitude are associated with darker colors; dark-is-more bias). See Schoenlein et al (2023) for a method to estimate combined merit from two sources, with potential to scale to include additional sources of merit that are known and those yet to be discovered. …”

More of what? Dissociating effects of conceptual and numeric mappings on interpreting colormap data visualizations

“…In colormap data visualizations, variations of color are used to represent variations in magnitude within a dataset. When observers interpret colormaps, they have expectations about how colors should map to magnitude (Cuff, 1973 ; McGranaghan, 1989 ; Schloss et al, 2019 ; Schoenlein et al, 2023 ; Sibrel et al, 2020 ), known as their inferred mappings. Interpreting colormaps, and information visualizations more broadly, is easier when visualization design matches people’s inferred mappings (Hegarty, 2011 ; Lin et al, 2013 ; Mukherjee et al, 2022 ; Norman, 2013 ; Schloss et al, 2018 , 2019 , 2021 ; Schoenlein et al, 2023 ; Sibrel et al, 2020 ; Tversky, 2011 ).…”