A structure-mapping function model for design-by-analogy: Knowledge representation and neurocognitive evaluation

Abstract

Design-by-analogy (DbA) enables the cross-domain transfer of design knowledge, thereby supporting the generation of novel product concepts. To facilitate such transfer, this study introduces a structure-mapping function model (SMFM) for representing design knowledge. SMFM captures two relational features—function and meta-function—aiming to support design knowledge transfer at these two in-depth levels. To evaluate its effectiveness, we conducted a design experiment involving human participants. Each participant used both SMFM and a classical function model, SAPPhIRE (State-Action-Part-Phenomenon-Input-oRgan-Effect), to solve two types of design tasks: function and meta-function problems. During the tasks, participants’ brain activity was recorded using Electroencephalography (EEG). Experimental findings are as follows: (1) In function problem solving, there were no significant differences between SMFM and SAPPhIRE in novelty (SMFM: M = 6.04, SD = 0.70; SAPPhIRE: M = 6.21, SD = 0.37; p > 0.05), quality (p > 0.05), or time spent (p > 0.05). (2) In meta-function problem solving, the SMFM group achieved significantly higher quality (M = 6.21 vs 5.92; p = 0.020) and exhibited stronger convergent thinking (p = 0.038), as indicated by increased activity in the frontocentral region of the left hemisphere. (3) The SAPPhIRE group spent more time on meta-function problems (M = 119.33s vs 100.35s; p = 0.036) and showed higher mental workload (p = 0.017), suggested by elevated centroparietal activity. These findings provide preliminary evidence that SMFM may better support meta-function problem solving in terms of solution quality and efficiency. The study also highlights cognitive differences associated with different function representations. A prototype retrieval tool based on SMFM has been developed to support analogy-driven conceptual design.