Design cognition differences when using unstructured, partially structured, and structured concept generation creativity techniques

This paper presents an experimental study of measuring and comparing design cognition while using different concept generation creativity techniques in a collaborative engineering design setting. Eleven design teams were given the same three design tasks, respectively, using an unstructured concept generation technique (brainstorming), a partially structured technique (morphological analysis), and a structured technique (TRIZ). A protocol analysis was carried out using the function–behavior–structure ontologically based coding scheme. This study showed that the students' design cognition was affected by the degree of structuredness of the concept generation creativity techniques they applied in their designing. The more structured a concept generation creativity technique is, the more likely that designers using this technique tend to focus more on problem-related aspects of designing, i.e., design goals and requirements. Results also indicate that the influence of the structuredness of concept generation creativity techniques mainly affects the early parts of the designing process, and differences between different design conditions exhibit an overall commonality of designing. Keywords: concept generation techniquedesign cognitionfunction–behavior–structure ontologyprotocol analysis Acknowledgements This research is supported by the National Science Foundation under Grant Nos NSF CMMI-0926908 and CMMI-1161715. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors thank Jacob Moore for his assistance in gathering study data, Sergei Chernyak and Matthew Dworsky for transcription and coding, and the study participants for their participation. Notes 1. As there is no nonparametric equivalent in Friedman's ANOVA, within-subject polynomial contrasts were still calculated in those sessions that did not meet the normality assumption.