ICONIC DISPLAYS AND INFORMATION COMMUNICATION: DISCRIMINATION AND LOCALIZATION OF CHANGES IN THE BAR, STAR, AND FACE

Three icons, the bar, star, and face were evaluated as predictor displays for use in process control. An icon calibration study was performed to equate icons and dimensions on a change discrimination task, where only one of the six total dimensions could change at any one time. The bar and star were easily equated, but the face never reached these high performance levels. In Experiment 1, the time between a target and test icon (ISI), number of dimensions changing, and amount of change was manipulated to assess interactions between variables and icons. Performance on the bar and star were approximately equal, followed by slightly poorer performance on the face when a single dimension changed. However, when several dimensions changed simultaneously, performance on all three icons was about equal. An analysis comparing d\u27 performance with that predicted by the normative model of the Theory of Signal Detectability revealed face performance levels above predicted for almost all conditions. Bar and star levels were almost always below predicted. In Experiment 2, the effect of ISI and probability that any single dimension could change, on a change localization task, was investigated. Overall, d\u27 performance on the face was best, followed by the bar and then the star. Subjects were able to divide their attention almost equally across all dimensions of the face. This was not true for the bar or star. For the tasks chosen here, the face was the most efficient transmitter of information when multiple dimensions changed simultaneously, even though it was less effective when a single dimension changed. It was hypothesized that the familiarity of the face enhanced the scanning, detection, and remembering of dimensions. Other icon studies were discussed in terms of the familiarity effect, and further research plans were discussed. Recommendations were made regarding the development and application of icons as predictor displays