Objective Design Fixation Analysis as a Basis for AI-Designer Teaming in Constrained Design

<p dir="ltr">For more than three decades, design fixation has been investigated by the design research community, mainly in the context of open-ended design tasks. Much of the work has focused on mitigating design fixation so that engineers and designers can fully explore a design space. Previous work subjectively evaluates the similarity of designers' solutions to example solutions and proposes myriad methods to mitigate fixation. However, not every design problem is open-ended, and engineering designers often face already well-defined problems. They also frequently use computer-based design software that constrains their design process. Therefore, open-ended design techniques to study and mitigate fixation, as they currently stand, are not well-suited to constrained design problems. Fortunately, constrained design problems lend themselves to the use of objective and quantitative metrics for detecting fixation, which in turn could enable the use of artificial intelligence (AI) based agents to guide designers during the design process. In addition, prior work has shown that neurocognitive signals measured using functional Near-Infrared Spectroscopy (fNIRS) have the potential to reveal designers' cognitive processes in a way that could be interpreted by an AI agent. However, this would require characterizing the relationship between those signals and objective fixation metrics. To address these challenges, this work presents a methodology to study and quantify design fixation in computer-based constrained design scenarios. The methodology is demonstrated using an adapted version of the HyForm Experimental Research Platform. Objective design fixation data and brain activation signals are collected and analyzed using quantitative metrics to understand how well participants explore the solution space. This research provides new tools for studying design fixation in iterative, constrained design tasks and paves the way for future development of online AI agents to assist designers.</p>