How Students Read Concept Maps: A Study of Eye Movements John C. Nesbit Faculty of Education, Simon Fraser University British Columbia, Canada nesbit@sfu.ca Hector Larios School of Interactive Arts and Technology, Simon Fraser University British Columbia, Canada hlarios@sfu.ca Olusola O. Adesope Faculty of Education, Simon Fraser University British Columbia, Canada ooa@sfu.ca Concept maps are network diagrams that show relationships among concepts. Despite their frequent use for instructional purposes, little is known about how learners read concept maps. The research reported here used eye-tracking equipment to investigate the order in which nodes are processed as learners begin reading a concept map. Three concept maps were presented to 36 undergraduate students. Each participant was individually tested in a one-hour session. Data analysis focused on the order of fixations in the first few seconds after a concept map was presented. The data showed that students began processing the concept maps by attending to nodes in the upper half of the diagrams, giving early attention to the center, upper- center, and upper-left regions. We interpret these results as showing the influence of two processing models, a text-transfer model that attempts to adapt the highly practiced order for reading text, and a hub-first model, a strategy that gives early attention to highly connected, superordinate information. Concept map designers can apply these findings by placing nodes containing superordinate information in upper-left and upper- center regions. Concept maps are spatial diagrams showing relationships among concepts as node-link assemblies. They show concepts as nodes, and relationships among the concepts as labeled links. Researchers' interest in the instructional use of concept maps has grown rapidly in recent years (Nesbit & Adesope, 2006; O’Donnell, Dansereau & Hall, 2002). A recent meta-analysis by Nesbit and Adesope (2006) found that studying with concept maps offered an average learning effect size of about .4 standard deviations when compared to studying with texts. The learning gain from concept maps is even more apparent when students construct their own concept maps or when they modify teachers’ maps. Despite the growing importance of concept maps as instructional tools, we have been unable to find research that collected eye movement data to investigate how learners read concept maps and how their graphical properties affect comprehension and learning. In this study we investigated the order in which nodes are viewed as learners begin interpreting the meaning of a concept map. Theoretical Framework The pace of eye movement research has quickened with the advent of technologies that allow precise, non-invasive measurement and recording (Rayner, 1998). There has been growing interest in using eye movement data to research cognitive inferences in reading (both silent and oral), scene perception, visual search, human-computer interaction, music reading and other tasks. Eye movement data have provided evidence for cognitive theories about how people read texts (Morris, 1994; Rayner, 1998; Sereno & Rayner, 1992). They reveal the self-regulation of attention in response to factors such as