Abstract
Research on colormaps demonstrated that expectations of how visual features map onto concepts are governed by two biases: dark-is-more (darker colors map to larger quantities) and opaque-is-more (more opaque regions map to larger quantities) (Schloss et al., 2019). We investigated whether these biases generalize to another system: Venn diagrams. Venn diagrams represent properties with overlapping regions and express logical relationships between those properties by using lightness level to encode “non-existence.” Given that “non-existence” is the extreme opposite of “more,” the dark-is-more bias implies people will infer that lighter regions signify non-existence. The opaque-is-more bias implies people will infer the least opaque region signifies non-existence. At its extreme, the least opaque region can appear as transparent or as a hole. Holes in surfaces arise from non-existing material, so we predicted people would analogously infer that regions appearing as holes map to non-existence (hole hypothesis). We tested these predictions by presenting participants with a series of logical statements, each paired with two Venn diagrams. Their task was to indicate which diagram best matched the statement. We determined the proportion of diagram choices consistent with inferring that light or dark regions signified non-existence. We varied diagram and background color so as to dissociate effects of dark-is-more and opaque-is-more biases and test the hole hypothesis. Experiment 1 provided evidence especially for the opaque-is-more bias; responses suggested participants inferred the least opaque region mapped to non-existence (p<.001). Experiment 2 discounted an alternative lightness-contrast account; background lightness only mattered when color relations were consistent with opacity variation (p<.001). Experiment 3 supported the hole hypothesis; background lightness had a larger effect when there was perceptual evidence for a hole compared to when there was merely perceptual evidence for opacity variation (p=.001). Thus, previously observed biases generalize across different systems, and perceptual organization influences interpretations of Venn diagrams.