This paper describes a computational model of skilled use of a graphical user interface based on Kintsch's construction-integration theory [4, 8]. The model uses knowledge of a detailed representation of information on the display, a user's goals and expectations, knowledge about the interface, and knowledge about the application domain to compute actions necessary to accomplish the user's current goal. The model provides a well-motivated account of one kind of errors, action slips [14], made by skilled users. We show how information about the intermediate state of a task on the display plays a critical role in skilled performance, i.e., display-based problem solving [10].
Muneo Kitajima, Peter G. Polson