Comprehension of Complex Visual Displays 1 Running Head: COMPREHENSION OF COMPLEX VISUAL DISPLAYS Thinking about the Weather: How Display Salience and Knowledge Affect Performance in a Graphic Inference Task

Three experiments examined how bottom-up and top-down processes interact when people view and make inferences from complex visual displays (weather maps). Bottom-up effects of display design were investigated by manipulating the relative visual salience of task-relevant and task-irrelevant information across different maps. Top-down effects of domain knowledge were investigated by examining performance and eye fixations before and after participants learned relevant meteorological principles. Map design and knowledge interacted such that salience had no effect on performance before participants learned the meteorological principles, but after learning, participants were more accurate if they viewed maps that made task-relevant information more visually salient. Effects of display design on task performance were somewhat dissociated from effects of display design on eye fixations. The results support a model in which eye fixations are directed primarily by top-down factors (task and domain knowledge). They suggest that good display design facilitates performance not just by guiding where viewers look in a complex display, but also by facilitating processing of the visual features that represent task-relevant information at a given display location. Suppose you are about to travel to San Francisco and consult a weather map to decide which clothes to pack. When you first view the map, is your attention guided by knowledge of the location of San Francisco or is automatically grabbed by prominent features such as a bright red area of particularly hot temperature over Arizona? Recent developments in scientific and information visualization make it easy to create complex visual displays, such as weather maps, in which multiple visual variables are superimposed on the same space (Card, Mackinlay & Shneiderman, 1999; Thomas & Cook, 2005). Yet perhaps one of the most fundamental facts about human cognition is that attention is capacity limited, such that objects in the world, or in a visual display, compete for our attention (e.g., Desimone & Duncan, 1995). In this situation, how is the selection of goal-relevant information influenced conjointly by the design of the display and by knowledge? In this article, we examine effects of display design and domain knowledge on a graphical comprehension task in which people make inferences from maps. We demonstrate that display design has significant effects on graphical comprehension and show that display effects interact with knowledge. We examine patterns of task performance and eye fixations to elucidate models of how display design and knowledge interact in graphical comprehension. Our research is informed by, yet differs …