The primary research objective of this project was to gain a scientific understanding of how individuals make design process-related decisions, specifically information acquisition decisions. The approach involved a synthesis of descriptive theories from psychological and cognitive science, and empirical evidence from individual decisions within different design situations. The project team took a broad view of information acquisition activity to include modeling & simulation, and communication with team members to acquire necessary information. Various factors that affect individuals’ information acquisition decisions, including prior knowledge, design dependencies, and the state of the design were considered.

Specifically, the project conducted studies on (i) individuals’ behaviors in parametric design tasks, where multiple noisy information sources are present and the total budget for information acquisition is limited, (ii) how knowledge influences individuals’ information acquisition decisions in engineering optimization tasks in the presence of constraints, (iii) how students on design teams acquire information and make decisions during open-ended engineering design projects, (iv) how inductive biases of the individuals influence information acquisition decisions, (v) how individuals make decisions to stop acquiring information in a competitive design scenario, (vi) how communication in design teams is used as a means for information acquisition in engineering design, and (vii) how state-of-the-art AI/ML techniques can be used for analyzing communication patterns in design teams. The studies resulted in design of behavioral experiments, observational data from the experiments and other observational studies, computational models of human behaviors, and behavioral insights that can be used to improve the design process outcomes.

The project has resulted in a mathematical framework for modeling the design process considering information acquisition decisions and sequential updating of belief as building blocks. The mathematical framework is demonstrated to be useful in conjunction with data from behavioral experiments and observational studies. This framework is useful for predicting designers’ performance in diverse design situations where sequential information acquisition is crucial, e.g. design of expensive experiments with limited budget. Using this framework, product development managers and systems engineers can predict design performance in terms of designers' decision strategies, their prior knowledge, and their information exchange strategies. The applications of this framework include design within organizations, and across organizational boundaries (e.g., in design crowdsourcing). The framework can also be used to develop agent-based models of engineering systems design.

The studies conducted in this project have resulted in a better understanding of strategies commonly used by individuals to make information acquisition decisions in design. The analysis of inductive biases has resulted in a better understanding of how designers use their prior knowledge as well as problem-specific aspects such as the range of the design space to make information acquisition decisions when uncertainty is high and not enough information is available through experimentation. The project has resulted in a better understanding of cognitive factors that influence an individual’s preferences leading to information acquisition decisions. With the combination of the descriptive models and statistical Bayesian inference, the quantification of individuals’ decision strategies for information acquisition is more streamlined.

Broader impacts:

The project has supported the training and education of five graduate students. The project provided unique training to the graduate students at the interface of engineering design, behavioral sciences, and cognitive sciences. Such training is currently unavailable in existing curricula. The students learnt the systematic approach towards designing lab experiments and using advanced machine learning techniques to develop data-driven models of behavioral phenomena. Several undergraduate students were engaged in the project activities through their participation in behavioral experiments. The results of this project have been incorporated into a graduate-level engineering course to expose students to interdisciplinary research on systems science with an emphasis on design decision making.

Understanding how humans make information acquisition decisions, and its impact on design outcomes, is crucial for improving engineering design and systems engineering. This understanding helps in developing more realistic representations of human behaviors in computational models for systems engineering and design, and emerging design scenarios. Improved computational models are essential for designing large-scale complex systems. The project enhanced collaboration between systems design and social science researchers, and facilitated the integration of theories and research methods from multiple disciplines. The results of the project have been disseminated through several journal publications and peer-reviewed conference papers in engineering design, systems engineering, decision making, and cognitive science communities.

Last Modified: 11/28/2022
Modified by: Jitesh H Panchal