ABSTRACT
Analogue representations are usually best for the design of interfaces in perceptualmotor tasks such as car driving. Many computer-based tasks, though, are carried out by manipulating symbols rather than performing skills. Skilled performance is graded – we can talk about the performance of task elements (such as operating the controls, performing hill starts and reacting to traffic signals) in terms of the degree of success or safety with which they were carried out. Performance can be assessed and scored on a continuum or scale. Symbol-based human-computer interaction is much more brittle. Symbols always have a certain arbitrariness about them that increases the requirement for learning and the likelihood of forgetting. When recognising a symbol, we are usually either right or wrong – there are no shades of grey.
