What’s wrong with the rainbow? An interdisciplinary review of empirical evidence for and against the rainbow color scheme in visualizations

Rainbow color scheme is popularly used across the board in many scientific disciplines for visualizing data, yet heavily debated in visualization literature. In this paper, we first report an interdisciplinary visualization survey examining the prevalence rainbow color scheme use in scientific publications including a temporal analysis. Then we consolidate findings from empirical studies on the subject to better understand why and precisely when the rainbow color scheme might impair human performance with visualizations. Consequently, we systematically document and analyze the consequences of using the rainbow color scheme based on over 37,000 figures in 11,808 papers in information visualization, neuroscience, hydrology, geography, remote sensing and planetary science. Our findings reveal that while the rainbow scheme appears less and less in visualization-related publications, it remains popular in other scientific domains including remote sensing and planetary sciences (strikingly fourteen times more frequent in remote sensing and planetary sciences than in information visualization outlets). We also find that conflicting findings about human performance with the rainbow color scheme is most likely explained by what the users are asked to do with it (i.e., task type). We detail and typify tasks used in the related empirical work in an effort to organize the current understanding on the subject, and translate it to practicable recommendations. We believe our review and analyses bring clarity and nuance to “the rainbow debate” and enable better-informed visualization advice.