Kenneth Kotovsky - US grants

This small grant for exploratory research (SGER), that is funded by Engineering Design and Cognitive Psychology, will begin to explore significant aspects of the cognitive basis for creative design. Initial work is based on a premise that one mechanism for creative design comes from the restriction of search in a problem space representation due to a constraint, known as an impasse. The impasse prevents search to a satisfactory solution, requiring a change in the representation that is used as a model of the design problem search space. The purpose of this seed grant is to investigate how changes in representation that enable the overcoming of impasses are generated. This one-year seed proposal provides initial work toward uncovering the potential interleaving of cognitive reasoning mechanisms for creative problem solving and innovative design processes. The long-term goal is to use this understanding to improve the innovation process through better tools and methods that stimulate human creativity, or provide a platform for computer design tools that advise the innovation process.
The intellectual merit of this work is a deeper understanding of how people overcome significant hurdles in creative problem solving -- how people find the aha that helps them solve difficult design problems that lead to new innovations. The broader impact of this work is the improvement in the efficiency and effectiveness in the innovation process by better understanding the mechanisms of thought that lead to innovations and developing tools to augment that process.
Two research tasks will be pursued in this seed grant. One is a cognitive science approach that seeks to uncover the basic mechanisms of overcoming impasses. The second seeks to uncover a model of how representation change is done during the design process. The basic investigation will be of how idea generation is affected by exposure to environmental stimuli (hints) and what the underlying cognitive control mechanisms are. The second research task will be to study how people change representation during the conceptual design process. This task will use protocol analysis and design traces to analyze what people's representation is at any given time during the design process, how it changes and why, and whether it iterates. This work will explore how representations are changed due to impasses. The expected outcome is to create a framework and the initial findings to pursue in a longer-term research project.

Most people have had the experience of ''getting stuck'' while trying to solve a hard problem. In the context of designing a new kind of low-emissions vehicle, for instance, a designer might reach an impasse in trying to figure out how to optimally position the fuel tank, given the many other constraints on vehicle design that must also be satisfied. Impasses in problem-solving can occur when the problem solver is fixated on a single representation of the problem space. Such impasses can be overcome only if one steps back and changes how the problem space is represented.

With support of the National Science Foundation, Drs. Cagan and Kotovsky are investigating cognitive and technological factors that influence how problem spaces in the domain of engineering design are represented, and how such representations can be changed in the face of an impasse. Their work promises to advance our understanding of how people overcome significant hurdles in creative problem solving -- how people find the ''aha'' that helps them solve difficult design problems that lead to new innovations. The project consists of laboratory work to collect experimental data on problem solving under impasse conditions, coupled with agent-based computational models of core cognitive processes involved in problem solving. An important goal of these research activities is to create knowledge that can be used in the classroom and workplace to more effectively stimulate human creativity, and provide a platform for computer design tools that advise and support the innovation process.

Innovation occurs within the minds of designers, within design teams and within computational design engines, each having functional similarities. This project studies the critical elements of creativity and innovation by formally understanding the role of individual cognition in group environments and its relation to computational simulations of the process to enhance the innovation process. An examination of these interactions reveals a great congruence in the structuring of these complex cognitive and social interactions, and the elucidation of the form of that structure and the dynamics of its operation into a model of the innovation process is the major goal of this project. This study could significantly influence how design innovation is structured and pursued in industrial settings. As such, this unique interdisciplinary research in the area of innovation is an appropriate match to the SciSIP mission.
This project rests on the assumption that an individual is working on a difficult design problem as part of a team effort. Often, iterative search within an individual mental representation makes improvements but yields no solution within that representation that meets the design goals. At that point the individual reaches an impasse or block. To overcome the impasse a new representation must emerge, one that occurs through search for a new representation coupled with inputs from the environment. At the individual level impasses can be overcome by appropriate stimuli at the appropriate time in the solution process, not too early and not too late.
The model posits that team members collaborate to develop a common representation within which a solution is found. The individual's representations influence other group members' representations and contribute to the overall group representation, and group discussions stimulate changes to both the individual and effective group representation, the group inputs acting as the external environmental stimuli to overcome the impasses mentioned above. Individuals then work with and develop their own representation, and at times collaborate with the group to both challenge and expand the group's representation as they search for solutions to the design task.
Software agents can emulate and support the group process, providing deeper understanding of the process and means to create a new type of design assist tools at the innovation level. The software agents use a given representation but, based on collaborative strategies, that representation can evolve as the agents work together as a team.
This study develops: (1) an understanding of the common structural and dynamic operations that occur throughout the design process; (2) a better understanding of how humans solve difficult design problems; (3) a deeper understanding of how people represent design problems and a resultant method for changing representations of a problem so as to find creative solutions; (4) a computer tools that build on these findings to improve search of a design problem; (5) an understanding of the relationship between individual problem solving and group problem solving; (6) a methodology and algorithm for group design processes; (7) new methods to assist humans in the creative process both as individuals and, through more efficaciously linking individuals within teams, as teams, so as to empower that process; and (8) the emergence of an empirically anchored theoretical model of individual and group cognition in design.
Formalizing a deeper understanding of how individuals participate and adapt in creative team problem solving, and how the team develops into a more efficient set of individuals is one broader outcome of this work. The cognitive understanding will inform the fundamental understanding of the cognition of creative problem solving and in particular design. Computational tools based on design agents can effectively generate innovative solutions.