Archives of Design Research

Conceptual Synthesis is a process by which two basic ideas are synthesized, which serves as a framework for new concept generation during the design process. It is the most effective process and is frequently employed in the workplace. According to Lubart (1994) and Rothenberg (1979), it is the simplest and most essential process for new concept generation, and Nagai (2009) found that it paralleled the Conceptual Combination Process in that it also represents a process by which basic concepts are synthesized to generate a complex concept.

The importance of concept combinations to creativity has been intensively studied by cognitive psychology research. Indeed, the concept combination process is a result of various human cognitive processes working together in synthesis. Therefore, exploring the mechanism by which these concepts are combined is critical to the generation of new ideas allowing the understanding and uncovering of human cognitive flexibility and productivity so as to expand existing knowledge systems. According to Wisniewski (1996), Lagne (2000), and Wisniewski & Love (1998), concept combination strategies include property mapping, concept blending, and relational linking. The creativity of the newly-created concept depends on the chosen strategy. Furthermore, our choice of strategy is determined by a variety of factors, such as the salience of concept features, and the similarity and abstraction level of combined concepts.

Nagai et al. (2009; 2005) found that the outputs produced from property mapping are limited in terms of their creativity as they are confined within the boundaries of the given concept. On the other hand, Nagai argued that concept blending generated more novel concepts as the newly-created concepts did not fall under the categories of either of the two original basic concepts. This finding echoes the results of Wilkenfeld and Ward (2001) and Wisniewski (1996; 1997), which found that people tended to utilize property mapping as their interpretation strategy when encountering a new word (noun) created by combining two separate nouns with highly similar concepts, while they tended to employ a relational linking strategy when the pair of nouns presented highly dissimilar concepts. In this context, the authors further asserted that relational thinking was the most effective strategy among the conceptual combination strategies, in that it produces the most creative output.

In combining concepts, the similarity between the two concepts can be seen to affect the creativity of the design product. The discussion structure can be organized as per Table 1.

Since design deals with the production and refinement of pictorial representations with the aim to generate new forms, it is suggested that designers make frequent use of visual thinking. According to Katz (1983), individuals with a natural ability to generate and manipulate mental images may prefer to use pictorial stimuli in problem-solving tasks.

This matches the general assumption (Hanington, 2003; Henderson, 1998; Muller, 1989; Tovey, 1992) that designers have a preference for the use of visual stimuli in the design process. Additionally, according to Sarkar & Chakrabarti (2008), images are considered to be the most effective representation modality for designers, which the latter opportunistically use at the idea generation stage (Casakin & Goldschmidt, 2000; Goldschmidt & Smolkov, 2006). This has been highlighted by Lugt’s study, in which the author examined whether sketches would affect idea generation as a visual stimulus during design group meetings. Lugt (2005) argued that sketching stimulates creativity by supporting the re-interpretation of an idea, providing new directions for the idea generation process, and facilitating a more integrated group process.

In this way, visual stimuli have played the role of most important information provider in design activity. Out of all types of visual stimuli, sketches have been dealt with most frequently in previous design research. However, designers form, comprehend, and transform images through new interpretations using not only sketches as a thinking tool but all the visual stimuli they encounter.

Despite the extensive use of visual displays in design practice, few empirical studies have been conducted to examine the way visual displays are utilized in design. In particular, there has been no research on the role of image stimuli in a designer’s conceptual synthesis process. Previous studies have offered results from experiments on the process of combining concepts using text. Therefore, this research aims to gain an insight into the thinking process of designers using images by enabling the designer to form new ideas using the two images presented to them.

The present study therefore sets out to answer the following question: What kind of impact does the similarity of images expressed in specific shapes have on the conception of design ideas? In other words, this paper aims to discover the relationship between the similarity of visual stimuli and the creativity produced by the visual stimuli in the conceptual synthesizing process, which represents the most frequently employed design idea generation technique.

Figure 1 Research Model

For the Conceptual Synthesis, we selected nouns presented in “A Standardized Set of 260 Picture for Conceptual Synthesis: Norms for Name Agreement, Image Agreement, Familiarity, and Visual Complexity” by Snodgrass and Vanderwart (1980), a standard picture set that is traditionally used in experimental psychology. We then grouped and selected the nouns according to the Battig and Montague Category Norms classification (Battig & Montague, 1969), as follows: ① Four-footed animal ② Kitchen utensil ③ Article of furniture ④ Part of human body ⑤ Fruit ⑥ Carpenter's tool ⑦ Articles of clothing ⑧ Part of building ⑨ Musical Instrument ⑩ Type of vehicle ⑪ Toy ⑫ Vegetable ⑬ Insect . We chose 7 nouns from each category to be utilized in the experiment. The selected examples from the 13 categories and their corresponding nouns are shown in Figure 1. As can be seen, the noun pairs from the same category and noun pairs from different categories were randomly selected and presented from a total of 91 nouns.

Figure 2 Examples of concept images

In order to compare the similarity and creativity of a text stimulus and image stimulus in a concept pair, one concept pair in each category was presented as a text stimulus as well as an image stimulus. This was conducted as a between-subjects experiment.

	1	2	3	4	5	6
Group A	…	Clock, Pumpkin (Different Category)			Spoon, Glass (Same Category)	…
Group B			Scissor, Pencil (Same Category)	Kettle, Grasshopper (Different Category)

The participants were 77 design freshmen. The 77 design freshmen were students learning basic design before pursuing a detailed major in design. They were relatively untrained in specific design idea conception techniques compared to students in other years, allowing us to observe the results in the case of ideas formed more naturally.

For the similarity assessment task (Screenshot C, D), the similarity of 40 randomly presented pairs of concepts was assessed on a 7-score scale. There was no time limit for the similarity rating. 5 minutes were given for the idea sketch task (Screenshot B). In the design idea sketch task, the same number of cases was presented before and after the similarity assessment.

Figure 3 Screenshots

The creativity of the design idea sketch was assessed with “the Korean version of the assessment tool for creative products” (Kim & Lee, 2004; Lee & Kim, 2007). The Korean version of the assessment tool for creative products was formed on the basis of numerous studies on the three-factor Creative Product Analysis Matrix by Besemer & Treffinger: Novelty, Resolution, and Elaboration and Synthesis. Besemer & Treffinger (1981) reviewed data from more than 90 research papers to draw more than 125 components of creativity. Following this, the author analyzed the similarities among these components and classified them, systematizing the evaluation criteria. The current structure results from over 20 years of continual validation and modification, and has proved applicable in many countries. The Korean version was modified so as to reflect the distinct characteristics of Korean culture in order to effectively assess the creativity of the outputs.

A team of three experts including two design professors and one PhD student subjectively evaluated the creativity of the idea sketches utilizing the Korean version of the assessment tool for creative products. All three assessors were design majors with at least a master’s degree in design and more than three years’ working experience in the field. The assessors did not consult one another in the course of their evaluations. The sketches for the same stimulus were shown to all three assessors simultaneously. They were first asked to relatively assess the sketches overall, and then to perform a specific assessment of the sketches in random order. The sketches were evaluated using the items presented above on a scale from 1 to 7, and the average of the scores from the three assessors was defined as the ‘creativity value’. The inter-rater reliability of the overall creativity assessments proved high, with an alpha coefficient of 0.635.

The objective of the present study was to probe the effects of the categories, representational modality, and level of similarity on creativity. A concept pair drawing a greater inter-participant variability in terms of creativity score than the average standard deviation for all the concept pairs meant that the participants’ creativity scores for this particular concept pair varied considerably. Hence, among the 23 randomly presented concept pairs, we eliminated the ones for which the inter-participant variability in the creativity scores was greater than the average standard deviation for all the concept pairs. Our final model consisted of a total of 12 concept pairs.

From our examination of the similarities and differences in creativity among the 12 selected concept pairs, we found that the mean similarity for items within the same category was 4.4, the mean similarity for items in different categories was 2.8, the mean difference test statistic was 7.897, and the corresponding p-value was lower than the significance level (a=0.05). Moreover, the difference tests for the creativity items and the means of creativity among various categories indicated a higher creativity score for items from different categories, which was found to be statistically significant.

Hypothesis 1. In design idea generation through conceptual combination, the creativity scores for idea sketches generated from concept pairs of low similarity is higher than the creativity scores for those generated from concept pairs of high similarity. In other words, a low similarity induces high creativity scores.

Examining the correlation between the similarity and creativity scores and the mean of creativity for the 12 concept pairs, we found that there was no correlation between similarity and novelty, similarity and resolution, or similarity and elaboration and synthesis. Therefore, hypothesis 1 was rejected: there was no correlation between similarity and creativity.

This led us to formulate a new hypothesis. The concept pairs in different categories had low similarities but high creativities. However, if no correlation existed between similarity and creativity across the concept pairs, we could assume that there would be high creativity when the similarity was low for concept pairs in the same category, which usually have high similarities. Likewise, we could also predict that there would be high creativity in cases of high similarity for concept pairs in different categories, which usually have low similarities.

Hypothesis 2-1. For concept pairs in the same category, the lower the similarity, the higher the creativity.

Hypothesis 2-2. For concept pairs in different categories, the higher the similar ity, the higher the creativity.

There was no correlation between similarity and creativity for concept pairs in the same category. Thus, hypothesis 2 was rejected. However, we found that there was a significant correlation between the similarity and the mean of creativity for concept pairs in different categories. Therefore, hypothesis 2-2 was supported. In other words, for concept pairs in different categories, the higher the similarity, the higher the creativity (see Figure 4.)

Figure 4 Idea sketch results according to perceived similarity and concept categories

Hypothesis 3-1. The creativity scores for design idea sketches inspired by concept pairs in the same category are higher when the stimuli are presented as images than when presented as text.

Hypothesis 3-2. The creativity scores for design idea sketches inspired by concept pairs in different categories are higher when the stimuli are presented as images than when presented as text.

For concept pairs in the same category, the difference test for the similarity and creativity relative to the presentational mode indicated a 0.1 point higher creativity score for image stimuli. The statistics for this difference were -0.046 with a p-value of 0.963, which was not statistically significant. Therefore, hypothesis 3-1 was rejected. However, the difference test for the similarity and creativity relative to different representational modalities for concept pairs in different categories indicated a 3.6 point higher creativity score for image stimuli. The statistics for this difference were -2.249 with a p-value of 0.026, which was statistically significant. Thus, hypothesis 3-2, which predicted a higher creativity for image stimuli, was supported.

Our hypothesis that creativity scores are higher for visual stimuli can therefore be supported (See Figure 5). Regarding similarity, there was no difference between the different modes of presentation. The hypothesis that the level of similarity is higher when concept pairs are presented as images must therefore be rejected.

Figure 5 Idea sketch results according to presentational mode and concept categories