Computational modeling for improving usability design workflow

Engineering practice moves slowly toward practical user centered product design for several reasons. Efficient tools are missing to handle complexity of the vast usability design space, in a stark contrast to a variety of CAD tools for other design disciplines. Designers still largely rely on ergonomic principles/heuristics. Consequently shortage of user centered perspective shows especially in early design stages. Empirically identifying usability problems later by human subjects in the prototyping stage, requires expensive user tests and design revisions. This paper explores how modern cognitive modeling techniques can systematically improve usability design to simultaneously address these challenges. A cognitive modeling aided design (CogMAD) paradigm is investigated with the different requirements in conceptual and detailed design stages, the procedural framework, and its special training and design modes. In the preliminary proof of concept study, cognitive models are used in conjunction with empirical experiments to examine various user interface designs. A group of human subjects are asked to test a set of GUI designs with different settings for three design variables. In parallel, cognitive models, using the ACT-R/PM architecture with diverse visual search strategies commonly employed by users, are developed to simulate these interactions. It shows that cognitive modeling has the potential to greatly assist practical decision-making in engineering product design. Cognitive models can pinpoint possible cognitive challenges when human users deal with product interfaces in order to seek better usability designs. This study is our first step toward a design paradigm to integrate computational cognitive modeling tools for practical usability design.