| In this dissertation, we study the relationship between eye movements and visual attention. Different types of eye movements are investigated including microsaccades, eye fixation, and eye pursuit. First, we demonstrate that microsaccades occur during pursuit and they are linked to covert attention shifts. Employing a psychophysical task that involves covert attention shifts to a peripheral square, we detect if microsaccades occur during eye pursuit, and, if so, whether, and in what way, microsaccades are related to attention shifts. Microsaccades are found to occur during pursuit and they present in similar patterns as those occurring during eye fixation. We discover that microsaccades tend to be biased towards the same direction as pursuit and the bias increases with increases of pursuit velocities. Through the analysis of microsaccade orientation and latency, we argue that microsaccades occurring during pursuit, rather than being randomly distributed, have a link with covert attention shifts. This is consistent with what has been reported for microsaccades occurring during fixation. Further analysis of microsaccade amplitude supports this argument. The potential attention mechanisms for the characteristics of microsaccades are discussed. We suggest that the attention allocation during pursuit is responsible for the characteristics of microsaccades. Our analyses of microsaccades also enforce the argument that microsaccades may be the suppressed saccades.;In addition to microsaccades, the attention allocation during eye fixation and eye pursuit are considered as well. We demonstrate that, during eye fixation, the local image content around the area of a fixation point is a significant factor to influence the fixation duration. However, during pursuit, the pursuit direction, rather than image contents, is important to decide attention allocation. According to these results, a top-down attention model based on types of eye movements is built. Three types of eye movements are considered separately in the model. They are eye fixation, eye pursuit, and saccadic eye movements (including microsaccades). The model is applied to the design of an interactive 2D video game. We demonstrate that the game is successfully designed in different difficulty levels through the analysis of attention allocation by our attention model. Our results imply that the attention modeling can be used to alter the game play so as to provide varying difficulty levels and it is also promising to take advantage of eye tracking data for broader applications, such as for navigation, intelligent map searching, augmented reality, and others. |
