Finding out what users really think: using cognitive modelling to improve icon design
Ratings were obtained from users of the perceived relationships between problems users were required to solve on an interface and the solutions which were offered by one of 3 icon sets. Pathfinder analysis was used to access the structure of users5 ‘mental maps5 of the system. Analysis of relatedness ratings revealed that those using pictorial and abstract icon sets were able to learn iconproblem relationships quickly. However, further analysis revealed that those using abstract icon sets appeared to have a ‘fuzzier5, less definitive, conception of the interface. The implications for icon design are discussed.