Topics we're working on:

Achieving a Deeper Understanding of Design Teams through Computational Modeling

A team is more than the sum of the individuals of which it is composed, having the potential to become more effective through interaction between the constituent members. With respect to design, this benefit generally arises from the ability of the team members to explore a variety of design alternatives, but also to collaboratively focus on a shrinking set of the most promising alternatives. Although most engineering design occurs within teams, much cognitive research in engineering design has focused on individuals.

Empirical studies are one of the most common means for evaluating both individual and team-based design methods. However, these studies can incur a high personnel cost while only returning a limited amount of data. It can also be difficult to isolate the effects of specific characteristics. A computational model of design teams could augment traditional methods of investigation by enabling the rapid and resource-efficient evaluation of design strategies, and making it feasible to isolate specific cognitive phenomena.

This work developed the Cognitively Inspired Simulated Annealing Teams (CISAT) modeling framework (presented graphically below). CISAT is anagent-based platform that has been designed to simulate both the process and performance of human design teams. The basic functionality of individual agents is provided through simulated annealing constructs. A number of cognitive phenomena are also modeled in CISAT to provide a rich representation of team activity.

The CISAT modeling framework was used to recreate the results of a study that tasked small teams of engineers with designing a truss structure. While simulating the results of the study, CISAT was given access to the same objectives, operations, and feedback that was available to human teams in the original study. Comparing the simulated teams to the human teams indicated that CISAT reproduced many of the same trends that were observed in the original study (see below). Deeper analysis with CISAT also indicated that individuals' self-bias (preference for their own designs) may be a beneficial characteristic, helping to avoid premature convergence.

Future work will seek to implement more detailed models for learning and heuristic development within the CISAT framework, allowing agents to more intelligently sequence the application of move operators. The CISAT framework will also be used to model more complex design problems to provide opportunities for further validation and refinement.

Primary Researcher: Chris McComb

Using Neuroimaging to Understand Multi-Attribute Product Preference Judgments Involving Sustainability

Using user data to accurately collect, model, and predict consumer preferences continues to be a critical part of the engineering design process. To understand consumer behavior, engineering design teams utilize a wide array of both qualitative and quantitative methods. However, despite their power, both qualitative and quantitative research methods are ultimately limited by the fact that they all rely on direct input from the user themselves. This can be problematic due to the fact that potential users may not accurately represent their own true preferences. Furthermore, these individuals may be unable to express what they are truly thinking, feeling, or desiring.

Some preference judgments are particularly difficult to obtain accurate data for. For example, this is evident when dealing with preference judgments that involve difficult choices, such as sustainability. Sustainable products have largely underperformed in the consumer marketplace, and the reasons for this are unclear, especially due to the fact that consumer studies have shown that individuals desire environmentally smart products. Previously, work in our research group from Goucher-Lambert and Cagan showed that when a product is evaluated with its environmental impact present, users tend to perceive functional attributes to be more important, and aesthetics to be less important.

One way that researchers are exploring complex decision-making scenarios at the time of judgment is through the use of neuroimaging techniques. These methods provide a powerful set of tools to capture the neural processes underlying brain functions—including choice decisions. In our group we have begun exploring functional magnetic resonance imaging (fMRI) to explore engineering design research questions. This project seeks to utilize fMRI to understand the brain functions, and areas of brain activation associated with multi-attribute preference decisions involving sustainability.

Primary Researcher: Kosa Goucher-Lambert

The Role of Social Choice Theory and Arrow's Theorem in Engineering Design

Much of the design process is accomplished by teams rather than individuals. During design, there often arise situations in which members of a team have different opinions, yet a group decision must still be made. Unfortunately, Arrow's Impossibility Theorem indicates that there is no method for aggregating group preferences that will always satisfy a small number of "fair" conditions.

A broad debate within the engineering design literature has attempted to assess whether or not Arrow's theorem applies to engineering design. Some researchers have proposed that it applies to all problem with multiple criteria or multiple decision-makers. Others have adopted the stance that engineering design is primarily concerned with the aggregation of multiple criteria, and thus Arrow's theorem need not apply. This project took the point of view that Arrow's theorem only applies to early stages of design (when criteria for comparing solutions are not well-defined).

With this in mind, we studied which voting rules were more likely to lead to outcomes that satisfied Arrow's conditions during the early stages of design. To do this, we first used experiential conjoint methodology to query real preferences for a set of 3D printed mugs (see above). We then used those preferences to construct, simulate, and analyze thousands of voting scenarios with different numbers of alternatives and voters, and several different voting rules.

Results indicated that the Copeland voting rule (shown above) offered the highest probability of satisfying all of Arrow's conditions. In addition, the Copeland rule was the most strategyproof (most resistant to manipulation by a dishonest individual). A comparison of the results for different voting rules is shown below.

Future work will extend this analysis to a larger set of aggregation functions, and explore the use of the Copeland function in more complex and longitudinal design contexts.

Researchers: Kosa Goucher-Lambert & Chris McComb

Merging Computational Design & Biology

Jon Cagan and Phil LeDuc talk about their collaborative work bridging the gap between computational design and biomechanics.