Quantifying Designer and User Factors in Engineering Design Using Psychophysiological Measurements and Human Responses

Wan-Lin Hu, Purdue University

Abstract

This research considers both the designer and the user side of engineering design, and proposes an interdisciplinary approach to conduct quantitative and qualitative analyses of the effects of cognitive states and human characteristics on the relationships between humans and design outcomes. Design outcomes studied in this research include quantity, quality, novelty, elaboration, liking, goodness, and trustworthiness. The first goal of this research is to assess the effects of cognitive states, demographic factors, and personal experience of designers on the engineering design process and outcomes, and identify sets of features that can contribute to successful design outcomes. The second goal is to characterize the influences of psychophysiological responses and demographics of users on their perception of design outcomes, primarily trust. These two research goals have been achieved with four individual studies. Study 1 established a support vector machine (SVM)-based prediction model to characterize the relationship between the novelty, quality, and elaboration of design concepts and the EEG metrics as well as some demographic factors of designers. Results characterize the combination of engagement and workload that is correlated with good design outcomes. For example, distraction is positively correlated with design novelty, and highly active attention is correlated with good design quality. Study 2 used empirical evidence and quantitative methods to show the effects of novice designers' contextual experience and demographic background on design tasks, particularly as they relate to the design process and design outcomes. Results suggest that contextual experience is negatively correlated with distraction during ideation and the novelty of proposed solutions. Study 3 described an empirical trust sensor model that maps psychophysiological measurements of users to their perceived trustworthiness of an engineering system. Several EEG and GSR features were identified as predictors to changes in trust level. A mean accuracy of 71.57% is achieved using a combination of classifiers to model trust level in each human subject. Study 4 developed a quantitative model to describe users' perceived trustworthiness of an engineering system based on their experience, expectation bias, and cumulative trust. This model can incorporate effects of system error types and user characteristics including cultural background and gender. The goodness of fit exceeds 91% for the general population as verified using data collected from over 900 participants. The proposed approach and results have broad implications for design methodology development and engineering system design and control. On the designer side, this approach provides a direct and objective assessment of designers' cognitive states and the effect of design methodology and/or interventions. The prediction model can further use psychophysiological measurements along with demographic factors to partially replace or augment traditional ideation metrics and to improve the efficacy of ideation research. Moreover, the gained knowledge of the influences of designer characteristics, including personal experience and demographic background, on ideation process and design outcomes can facilitate the development of design methodology for different groups, especially for novice designers. From a user-centered perspective, the proposed approach can enable engineering systems to sense user trust level based on their cognitive states and behaviors. The quantitative model for describing the influences of user characteristics and demographics further enhances the systems to respond to different groups adequately in the era of globalization. Most importantly, these models will allow engineering systems to respond to trust behaviors in order to achieve a successful interaction between humans and engineering systems.

Degree

Ph.D.

Advisors

Reid, Purdue University.

Subject Area

Mechanical engineering

COinS