ABSTRACT
When the elements of a display and their controls vary along only one dimension, most people would predict that performance would be best if the controls were mapped to the display elements in a manner that maintains spatial correspondence (Vu & Proctor, 2003). However, for many interfaces, the displays and controls vary along two or more dimensions. Predicting performance with various mappings for these display-control configurations is often more difficult than one might assume. For multidimensional S-R sets, issues of compatibility become more complex because performance may depend on which dimension is defined as relevant for the task, structural relations between the stimulus and response sets play an increasing role, and compatibility may be more important for some dimensions than for others.
Compatibility effects occur when the stimulus and response configurations can be classified as two dimensional, even for situations in which the stimulus dimension bears no resemblance to the response dimension. This point was first illustrated in a study by Fitts and Biederman (1965), following up on a finding reported by Morin, Forrin, and Archer (1961). Morin et al. used a task in which four two-dimensional stimuli (one or two squares or circles) were mapped to four keypresses made by the fingers on the right hand. The left-to-right mapping of stimuli to responses was single circle to index finger, single square to middle finger, two circles to ring finger, and two squares to the pinkie finger (see Figure 5.1). Morin et al. found that RT was 190 ms longer for this four-choice task than for a two-choice task in which a circle was mapped to the right index finger and a square to the right middle finger.
