More generally, Information Visualization is a way to reveal data properties which would not be trivially detected otherwise, to shed light on breakthroughs, and to share the poignant experience of “Aha, I see!” (Few 2006) thanks to its intuitive aspect. This research field contributes to the emergence of novel scientific theories by improving the exploitation of human cognition. According to Card, Mackinlay and Shneiderman (Card 1999), the main focus of visualization is indeed to amplify cognition. The authors listed a number of key ways to do so, showing the advantages of using visualization techniques during data exploration:
- Reducing time spent looking for information,
- Enhancing the recognition of patterns,
- Enabling perceptual inference operations,
- Using perceptual attention mechanisms for monitoring tasks,
- Encoding information in an actionable medium.
In this blog post we provide a short introduction to the perceptual support of visualization.
Information Visualization relies on the properties and perception abilities of the human visual system. According to Information Theory, vision is the sense that has the largest bandwidth (100 Mbits/s), which makes it the best suited channel to convey information to the brain (in contrast, audition has only around 100 bits/s) (Ware 2004). Visualization hence requires building and applying a visual language to encode information that can be read and interpreted correctly. This operation is called a mapping between data variables and visual variables. This language relies on visual features like geometric primitives, colors and sizes, and was theorized in (Bertin 1967) and (Cleveland 1984), and extended in (Mackinlay 1986).
However selecting visual features to convey information is not trivial. One would indeed like to select the most effective ones, but while avoiding misunderstandings and over-interpretations. Well-established guidelines distinct two kinds of data variables: quantitative and qualitative variables (see the following table). Visual features can be selected according to the type of data, but difficulties remain when mixing different visual variables in the same image.