Abstract
Fitting machines to their users is the goal of human factors. This task must be accomplished at both the physical interface and the cognitive interface. While the methods of physical interface design are fairly standard, few techniques are recognized for designing the cognitive interface. This paper presents and validates one method of cognitive interface design. Expert subjects provided similarity ratings for all possible pairs of machine functions. These similarity data represent psychological distances among functions. The distance data were subjected to a Multi-dimensional Scaling procedure to yield coordinates for each function in N-dimensional space. The coordinates were converted into euclidean distances for analysis by a hierarchical clustering algorithm. The desired number of “clustered functions” can then be selected for the system of interest. This procedure was employed to construct a menu of word processing functions. A paired-associate learning study demonstrated that significantly fewer errors were made in learning the menu organized via this procedure than were committed on a menu exhibiting a random structure.
Get full access to this article
View all access options for this article.