1 Visualization and HCI: A brief survey Draft • 3 Sept 1996 Jonas Löwgren Department of Computer and Information Science Linköping University, 581 83 Linköping, Sweden Tel +46 13 281482 • Fax +46 13 142231 • Email jlo@ida.liu.se 1. INTRODUCTION The concept of visualization essentially refers to the use of visual or graphical techniques for presenting information. In computing, there is typically also a connotation of large amounts of information being presented in a more compact way through the mapping to visual dimensions. Visualization in computing is a large, rapidly growing and heterogeneous field and this presentation is not aimed at providing a complete survey. Rather, I concentrate of the subfield that I find most appropriate for the Supreme project: interactive visualizations of abstract data. This means that topics such as scientific visualization, where much of the work concerns non-interactive visualization of large sets of concrete data, are disregarded. I have mainly used literature from the field of human-computer interaction (HCI), where the inter- est in visualization to support knowledge work is considerable. 2. WHY VISUALIZATION? Several arguments have been put forward in favor of visualization in computing. The general trend towards more visual user interfaces is well-known; on a more detailed level, the litera- ture offers fairly strong support in several areas. 2.1 HOLISTIC PERCEPTION In the Gestalt-psychological view of visual perception, the whole visualized structure is per- ceived as one meaningful unit (on some level of granularity) which is different from the sum of the details in the picture, and provides a frame of interpretation for them. Also, the inter- pretation of a picture is not a simple addition of piecewise interpretations. More holistic effects come into play (see Arnheim, 1956 for a treatment of Gestalt psychology as applied to the visual arts). The main argument in favor of visualization here is that by looking at a (Gestalt) view of a whole information space, the user learns more about its structure and is better equipped to plan local actions and assess their outcomes. Serendipity (finding useful information without looking for it) is another favorable effect of providing the view of the whole. 2.2 EXPERIMENTAL EVIDENCE Healey et al (1996) explore the use of preattentive processing, i.e., perception-based processing occuring before any conscious reasoning is possible, for the task of numeric estimation. They show that preattentive processing of a visualization based on hue and orientation works well, and also provide a more complete list of possible visual attributes: orientation, length, width, size, curvature, number, terminators, intersection, closure, hue, intensity, flicker, direction of motion, binocular luster, stereoscopic depth, 3D depth cues, lighting direction.