I'm an engineering student at the moment, but I've had a few ideas kicking around that might help you out, or give you a seed for another idea.
My first experience with the design process was interesting. In engineering, the standard is to take one concept to completion, changing it along the way as issues come up. It was immediately obvious that the multiple concept process helped to generate more innovative ideas. However, there were situations in which I thought that it could have helped if a bit more logic was infused into the process.
For example, selecting materials. My concept of "every material possible" was: I'll get a sample of a representative thermoplastic, thermoset, foamed thermoplastic, foamed thermoset, leather and fabric. That was pretty much every scientific material class that could be used to make what we were looking at. However, I was told that they wanted to see different colours and durometers of everything as well, then just pick one and that would be it. Obviously it's impossible to grab everything you could make something out of, and my goal was to go extremely broad, pick a category, pick a sub-category, etc. until the best possible solution was found. I still feel like this could have produced a better result.
The relevant study that I could see would be to have a really basic brief, and ask a variety of different professions to design for it. Obviously their results won't be as great, but it could provide some insight into how people with different backgrounds approach a problem, which may help to improve the design process.
During the Apple/Samsung lawsuit, a common line that was repeated was that Apple was patenting "a rectangle with rounded corners". So I did a quick calculation out of interest to see how many possibilities there were on a rectangle based on just edge shape with the options of corner, small round, large round (12 edges total). This gives 3^12 options for a total of about half a million. This gets reduced when factoring in symmetry and other restrictions, but it wouldn't be difficult to write a program to generate forms based on those parameters and restrictions that would generate all of the forms and spit out renderings into a pdf that the designers could pin up and choose from. If they notice that, for example, none of the ones that have large rounds on the front edge look good, they can tell the program to add that restriction and move one. This would allow them to reach the epitome of rounded rectangles because they have literally been through every one of the half million options in some way or another. Finishes and material combinations could also be included in the generation parameters.
For the study, build a program like that, run a few design challenges with it, see if it works, or if the roboticism of it takes away somehow.