Emergent Features and Graphical Elements: Designing More Effective Configural Displays

Abstract

When performing tasks in complex, dynamic domains individuals must consider information regarding both high-level constraints (relationships among several variables, performance goals) and low-level data (the values of individual variables). Previous research has revealed mixed results concerning the effectiveness of configural displays in achieving these dual design goals. Two empirical studies were conducted to investigate these issues using a laboratory analogue of a complex, dynamic task modeled on a real-world domain. Performance with a configural display, which highlighted the low-level data, was compared with performance with a bar graph display. For the extraction of information about high-level constraints in a memory probe task, the configural display significantly increased accuracy with no cost in latency. For low-level data there were no differences in accuracy across the two display conditions, but there was a significant cost in latency with the configural display. However, this cost was dependent on both experience and system state. These results suggest that configural displays can be designed to support the extraction of both high-level constraints and low-level data in complex, dynamic domains. To support the extraction of information for high-level constraints, the emergent features produced by a configural display must reflect the critical data relationships that are present in the domain. To support the extraction of low-level data, the graphical elements of the display must be made more salient perceptually through a variety of techniques, including emphasis of scale, spatial separation, and color-coding.