Douglas, Sarah A.2023-06-202023-06-201988-07-12https://hdl.handle.net/1794/284266 pagesMost of the existing analytical descriptions of users characterize their performance as a function of the cognitive representation of the command sequences of the computer-based task (e.g. Anderson, Farrell, & Sauers, 1984; Card, Moran & Newell, 1983; Polson & Kieras, 1985; Norman, 1986). This is represented as goal-oriented schemata: procedures, plans, or production rules. Thus, the interface designer need only lay out the command sequences adequate to achieve a set of core tasks to make predictions about user behavior. These models are by and large restricted to descriptions of error-free, skilled (expert) performance or error-free learner subsets of expert knowledge. Attempts to extend these analytical models to accommodate learner error soon find themselves coping with the problems of prior knowledge (c.f. Douglas & Moran, 1983; Riley, 1986). That is, elements of performance that are independent of the computer task representation. Additionally, the existing models make no attempt to represent the ongoing interactive nature of human behavior at the interface. This problem of taking into account aspects of human performance at the interface which are independent of the task representation can be called the context problem. In the remainder of this paper I will attempt to delineate the nature of this problem by defining the notion of context, giving examples of context accommodation in interface design, and discussing the practical and theoretical problems that context creates for user models and interface design.enCreative Commons BY-NC-ND 4.0-USUNIXsyntaxlinguisticsArticle