An 'Active Vision' Computational Model of Visual Search for Human-Computer Interaction
Abstract
Visual search is an important part of human-computer interaction (HCI). The visual search processes that people use have a substantial effect on the time expended and likelihood of finding the information they seek. This work investigates visual search through computational cognitive modeling of empirical data. Computational cognitive modeling is a powerful methodology that uses computer simulation to capture, assert, record, and replay plausible sets of interactions among the many human processes at work during visual search. This work aims to provide a cognitive model of visual search that can be utilized by predictive interface analysis tools and to do so in a manner consistent with a comprehensive theory of human visual processing, namely active vision. The model accounts for the four questions of active vision: What can be perceived in a fixation? When do the eyes move? Where do the eyes move? What information is integrated between eye movements? The answers to these questions are important to both practitioners and researchers in HCI. This work presents a principled progression of the development of a computational model of active vision. Three sets of data were modeled in the EPIC (Executive Process-Interactive Control) cognitive architecture. This work extends the practice of computational cognitive modeling by (a) providing the first detailed instantiation of the theory of active vision in a computational framework and (b) informing the process of developing computational models through the use of eye movement data. This instantiation allows us to better understand how these visual search processes can be used computationally to predict people's visual search behavior. The development of a comprehensive model ultimately benefits HCI by giving researchers and practitioners a better understanding of how users visually interact with computers, and provides a foundation for tools to predict that interaction.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2009
- Accession Number
- ADA518836
Entities
People
- Anthony J. Hornof
- Tim Halverson
Organizations
- Air Force Research Laboratory