Zoe S. Howard¹ (), Anna Chinni¹, Karen B. Schloss¹; ¹University of Wisconsin-Madison

To create easily interpretable information visualizations, it is useful to understand people’s expectations of how visual features (e.g., texture, color) map onto concepts. Prior work on color showed that observers use assignment inference to infer color-concept mappings that maximize the “goodness” of all color-concept pairs in the scope of the encoding system (global assignment), rather than mapping colors to their strongest associate (local assignment). Global assignment led participants to assign concepts to weakly associated colors, despite more strongly associated candidates (Schloss et al., 2018). Studies of assignment inference for colors focused only on the colors and concepts in the scope of the encoding system. However, this scope may be too narrow when considering perceptual features that can be highly representational, such as textures (e.g., brick textures specifically resemble bricks). When textures strongly evoke concepts outside of the encoding system, observers may be less open to assigning textures to weakly associated concepts, resulting in local over global assignment when the two conflict. To test this possibility, we had participants interpret bar charts in which two textured bars represented two concepts (“encoding system”). On each trial, they reported which bar corresponded to a “target” concept (non-prompted concepts were “non-targets”). We focused on trials where global vs. local assignment would produce opposite responses. Overall, assignments were no more likely to be global than local, but the probability of global assignment increased as globally optimal textures were less representational of concepts outside of the encoding system (p <.01), more associated with the target (p <.001), and less associated with the non-target (p <.001). These results expand our understanding of factors that contribute to assignment inference—when visual features are highly representational, it is necessary to account for concepts outside of the encoding system to predict how people ascribe meaning to visual features in information visualizations.

Acknowledgements: NSF award BCS-2419493 to K.B.S