Capturing the distinction between task and device errors in a formal model of user behaviour

In any complex interactive human-computer system, people are likely to make errors during its operation. In this paper, we describe a validation study of an existing generic model of user behaviour. The study is based on the data and conclusions from an independent prior experiment. We show that the current model does successfully capture the key concepts investigated in the experiment, particularly relating to results to do with the distinction between task and device-specific errors. However, we also highlight some apparent weaknesses in the current model with respect to initialisation errors, based on comparison with previously unpublished (and more detailed) data from the experiment. The differences between data and observed model behaviour suggest the need for new empirical research to determine what additional factors are at work. We also discuss the potential use of formal models of user behaviour in both informing, and generating further hypotheses about the causes of human error.