Studies focusing on two iconic attributes, form and relative position, suggested subjects rely on form to recognise a particular icon when the form is representational. However, when form is abstract users switch from attempting to use form, to using position. Results had suggested that some degree of positional learning still occurred for subjects using representational form. However, a more in depth study of this suggestion was impossible since the standard performance measures of error rate

more »

... nd reaction time settled quickly into an optimal level of performance. To gain better insight into what was occurring, cursor movements were tracked, allowing pre-emptive move data to be captured. The showed that, irrespective of the type of form presented, subject learned and eventually switched to relying on position to identify the icon. More importantly, results suggest that this learning continued after the traditional measures of reaction time and error rate had reached what Jordan et al. (1991) termed experienced user performance level, or asymptote. The suggestion therefore is that traditional performance measures are perhaps not sensitive enough to measure learning at the experienced end of the learning curve, and that interface designers should thus consider ways in which to capture this potentially enlightening information which, as yet, remains unexamined.