Friday, February 17, 2017
Dream team: CMU engineering and psychology researchers develop model to study design teams
While zero team interaction might not be the best option for every problem, researchers found that sometimes, it's best to work alone.
If you’ve ever worked on a team project, you know that a strong team will help a project soar—but a bad team can bring the whole project crashing to the ground.
Today, almost all the products you love—your car, your iPhone, your air conditioner—were conceptualized by a team of designers. A recent study conducted by Carnegie Mellon University collaborators Christopher McComb, Jonathan Cagan, and Kenneth Kotovsky sought to answer an important question for the design industry—how do you best design your design team?
According to the study, there are some problems where the most efficient team might be a team in which the members don’t interact with each other at all.
This was a surprising conclusion for the researchers, given the common assumption that teams are generally more effective than individuals.
“In earlier work, we looked at how individuals solve problems. Based on that understanding, we are now looking at when you take those individuals and put them together into a team, what happens? How do teams work?” explains Cagan, a professor in the Department of Mechanical Engineering.
“Specifically, we were looking at the number of people that should be on a team and how often they interact with each other and exchange information. Given a design problem, we looked at these characteristics to design the optimal team to solve this problem.”
The research team used the Cognitively-Inspired Simulated Annealing Teams (CISAT) framework, a computational model that can predict the performance of human design teams. The team ran more than 100,000 different problems with various team sizes and interaction frequencies to determine which combination leads to the best performance or solution for a given problem.
“The idea is that teams aren’t a one-size-fits-all thing. Depending on the problem you’re trying to solve, you may want your team to look very different,” says McComb, lead author and postdoctoral researcher in the Department of Mechanical Engineering.
McComb says when they applied their approach to a new problem, that of designing the layout of Internet of Things cooling systems, the CISAT model predicted that zero interaction between team members would lead to the best outcome. Namely, each person solves the problem and then selects the best of the set of answers.
To make sure the CISAT predictions held true for real people, the researchers ran a study with mechanical engineering seniors at Carnegie Mellon. The seniors were asked to solve a design problem that involved smart air coolers, smart sensors and processors.
“When we actually ran this study with mechanical engineering seniors, it showed the same results as the prediction,” says McComb. “We had some students who interacted very frequently, some who interacted moderately frequently, and some who just didn’t talk to each other. And the ones who didn’t talk to each other had the best performance on the problem.”
While zero team interaction might not be the best option for every problem, the study is an important step in understanding how design teams really function. The study also has larger implications for how we solve problems in general.
“People solve problems all the time in teams. What we’re showing is that you need to design your team for the problem, and we have a way to do that. We have a computational model that is able to simulate the properties of humans and the steps they go through to solve a variety of problems,” says Kotovsky, a professor in the Department of Psychology.
The paper lays the groundwork for further studies on team-based design methodology. “This is a good place to start, and there are a lot of interesting directions to go with this research,” says Cagan. "Although our focus was on design teams, the lessons learned have potential implications across all types of problem solving teams, for design is a type of problem solving. Exploring this broader impact is one area of our future work."
Jonathan Cagan is the George Tallman and Florence Barrett Ladd Professor in Mechanical Engineering, Associate Dean for Strategic Initiatives for the College of Engineering, and Co-Director of the Integrated Innovation Institute.
by Catherine Graham
McComb C, Cagan J, Kotovsky K. Optimizing Design Teams Based on Problem Properties: Computational Team Simulations and an Applied Empirical Test. ASME. J. Mech. Des. 2017;139(4):041101-041101-12. doi:10.1115/1.4035793.