The Difference of In-depth Level of Cognition at the Early Stage of Idea Generation
Deny Willy, Yukari Nagai, Muhammad Ihsan, Georgi V. Georgiev
This paper investigates the in-depth level of cognition at the early stage of idea generation between craftsmen and designers. Examining the early stage of idea generation may explain the fundamental thoughts in observing and defining design problems. In order to capture the in-depth level of cognition, we conduct an experiment of think aloud protocol, where verbalized thoughts were analyzed using concept network based on the associative concept analysis. Furthermore, the obtained associative concept were analyzed to identify its semantic relation based on Factor Analysis. The findings show craftsmen tend to activate low-weighted associative concept at in-depth level of cognition with smaller number of polysemous features that explain their concerns on tangible-related issues such as proportion, shape. While designers activate more highly-weighted associative concept with higher polysemous features concern on intangible-related issues such as surroundings-context (i.e. eating culture) user’s affective preferences (i.e. companion, appealingness).
Keywords: In-depth level of cognition, Early stage of idea generation, Designer, Craftsman.
This study focuses at the early stage of idea generation to capture the associative concept at in-depth level of cognition between craftsmen and designers. Examining the early stage of idea generation may explain the fundamental thoughts in observing and reframing design problems. Many attempts have been made to capture user’s affective preferences based on users as the subject. We examine the in-depth level of cognition of the creators (craftsmen and designers) that attempt to grasp user’s feeling in bringing successful impression of a product [1, 2]. In order to capture the in-depth level of cognition, we conduct an experiment of think aloud protocol, where verbalized thoughts were analyzed using concept network based on the associative concept analysis and semantic relation analysis.
1.1 Early Stage of Idea Generation
Idea generation which is consisted of define and ideate is the essential step in design thinking process. It is an interplay of cognitive and affective skills, a processes directed specifically to the resolution of a recognized difficulty . Following is 4 steps design thinking process, where the early stage of idea generation is mainly discussed in this paper:
(1) Imagination (the early stage of idea generation): A stage where to observe and reframe the design problem.
(2) Ideation (later stage of idea generation): Employ sketch, graph or paper model to generate conceptual ideas visually.
(3) Prototyping: A stage of making rough model to convey an idea concretely.
(4) Evaluation: A stage to acquire user’s feedback by evaluating affective preference
The next step after design thinking process is realization of the idea as the real production for commercial.
At the early stage of idea generation, generally known, one is under condition that more loose and free to imagine. This may maintain a fair stage in observing of craftsmen and designers first-hand experiences. The early stage of idea generation was associated with greater diversity of ideas , therefore it is reasonable to capture the fundamental of one’s thoughts.
1.2 In-depth Level of Cognition
Generally known, designers cannot explicitly express their thoughts, this latent sensitivity is widely researched in Cognitive Psychology. It is known as implicit cognition which is understood to be that which is not explicitly recognized or verbalized .
The explicit expression which is presumably a shallow analysis is referred to surface level of cognition, and underlying cognition which is difficult to express is referred in-depth level of cognition (feeling, taste, impression) [2, 6, 7]. Taura et al. (2010) explain implicit impressions could exist in the feelings and are implied underneath explicit impressions which is related to deep impressions . According to Nagai et al. (2011) humans establish extremely rich metaphorical concepts (within in-depth impression) as key features of cognition in creative design as a way of how a designer capture profound understanding of an object . Previous studies has focused capturing in-depth level of cognition (impressions) of the users and based on created artifact. Our study focus on the creators (craftsmen and designers) while enter a very vague area at the early stage of idea generation (imagination) (Fig.1)(Fig.2).
Figure 1 Focus of the research: In-depth level of cognition
1.3 Verbalized Thoughts and Associative Concept Analysis
To examine the structure of thoughts from subjective experiences, a think-aloud as a part of protocol analysis (Crutcher 1994) employs verbal reports of thinking (Ericsson and Simon, 1993). Subjects are instructed to describe their thoughts, how they observe and reframe the design problem through verbal expression. The verbalized thoughts would almost have to reflect some aspect of the regular cognitive processes . The verbalize thoughts will be modelled using computational model to reproduce the observable aspects of human thoughts (in-depth level of cognition).
The associative concept analysis is a procedure to capture associative concepts of an expression has to do with individual mental. The associative concept is comprised of six sub-types: connotative, collocative, social, affective, reflected and thematic (Mwihaki 2004). It is latent within in-depth level of cognition which difficult to expressed explicitly. Therefore, a concept network is suitable as an associative analysis tool for exploring the latent links among concepts. In the field of psychology, the concept networks depict human memory as an associative system, where a single idea can contain broad meaning (polysemous). The concept networks utilize a concept dictionary from The University of South Florida free association, rhyme, and word fragment norms database (USF norms database) [9, 10].
The aim of this research is to capture the difference of associative concept at in-depth level of cognition at the early stage of idea generation in design thinking process between craftsmen and designers. In order to be able to capture in-depth associative concept, we conduct an experimental study of think aloud protocol where designer and craftsman freely express their ideas verbally.
To capture the associative concept at in-depth level of cognition at the early stage of idea generation, we conduct an experimental study of think aloud protocol where craftsmen and designers were instructed to freely express their ideas verbally. Next, we use a concept network analysis based on the associative concept dictionary to extract the verbalized thought. Framework of this research comprised of following steps (Fig. 2):
Two craftsmen and two product designers conducted a think aloud protocol. They were requested to imagine designing a fruit basket/container and freely express their ideas verbally without necessary to draw or observe the object. The procedure was recorded as a verbal data, and the sorted verbal expressions were then transcribed into English. The verbal data which consisted of explicit words are transferred onto vector graphs (conceptual network on the basis of USF norms database) to obtain extraction of the highly-weighted associative words which indicated by the out-degree centrality score (ODC). Identify the difference of the concept network structures by analyzing:
a. The density of connection, it exhibits the property of idea in a form of associative concept network.
b. The semantic relation, to find the characteristic of the associative concepts at in-depth level of cognition using orthogonal semantic map based on Factor Analysis.
Four subjects (two Indonesian craftsmen and two Indonesian designers; age range, 27-51 years) participated in this experiment. Two designers are university graduate with experience in craft design and concern of natural material utilization. Two craftsmen gained their special expertises from generation, recognized as master craftsmen in traditional bamboo craft in local village.
The experiment was set up in a simple manner, both subjects, craftsmen and designers were not necessary to perform any physical activity such as drawing or observing stimuli. They deliberately conditioned in a minimum instruction to be able to capture the fundamental associative concepts. Such rigid instruction as determining of design theme, market segmentation, or design function is avoided, considering it may result an excessive information that unfair and mislead. Minimum instruction possibly maintain a fair stage in observing of craftsmen’s and designers’ first hand experience in observing and reframing the design problems.
The whole procedures were recorded as a verbal data and transcribed word by word. Following rules were employed, connecting words such as prepositions, a few general verbs, articles, and pronouns, also other less related explanations were omitted . Finally, the sorted verbal data consist only of nouns, adjectives, adverbs and verbs were translated into English and further were proceed into concept network analysis on the basis of USF norms database. (visualized as graphs using Pajek 2.05 with the algorithm of Fruchterman Reingold) .
4.3 Concept Network Analysis
At the first stage of analysis, there are 107 sorted verbal expressions obtained from craftsmen, and 102 sorted verbal expressions from designers. Craftsmen tend to express tangible aspects such as systems, material, production (bold text). Whilst, designers paid more attention on intangible-related issues such as user’s affective preferences and environmental context (bold text) (Table 1).
Table 1 Sorted verbal expressions
|Category||List of sorted verbal expressions (partly shown)|
|Craftsmen||capacity, dimension, measure, standard, super, big, count, size, leg, height, thin, shape, square, position, part, head, stack, body, solid, base, width, top, long, oval, three-dimensional, thick, centimeter, box, design, container, fruit, duck, salt, egg, adjust, buyer, function, capable, form, set, color, supply, bamboo, scar, spot, glue, mark, sandpaper, etc.|
|Designers||place, kitchen, pluck, tree, shop, sensation, reap, pick, preservation, tropical, rotten, fresh, delicious, interaction, inform, remind, children, invite, accommodate, people, way salad, commercial, habit, crowd, appeal, appreciate, attractive, dignity, snack, put, table, hang, fruit, wood, appear, stand, durian, banana, apple, orange, watermelon, grape, etc.|
The sorted verbal data further were proceed and visualized as graphs of concept network analysis (Fig. 3). Craftsmen’s concept networks generate 1941 vertices (nodes), and designers’ concept network generate 1662 vertices (nodes). The networks were too dense and complex for analysis, therefore a second stage is necessary to simplify the created networks by a reduction method. The systematic reduction is based on the considerations: Not all the words from verbalized protocols contribute to in-depth level of cognition, in addition, the number of surface level of cognition is overmuch. Following indicate numbers of low scores of explicit words/surface level of cognition (bold):
– Craftsmen (total: 1941 words): 0.000-0.010 = 1462 words; 0.020 = 352 words; 0.030 = 94 words; 0.040 = 27 words; 0.051 = 6 words.
– Designers (1662 words) : 0.000-0.010 = 1259 words; 0.021 = 293 words; 0.032=77 words; 0.043 = 23 words; 0.054 = 8 words; 0.065 = 1 word; 0.076 = 1 word.
Figure 3 Concept networks of craftsmen’s and designers’s in-depth level of cognition before reduction (words and scores not shown due to complexity)
5. Analysis and Result
5.1 Concept Network Analysis (after reduction)
The simplified concept reduce the words that play a less important role in the networks, so that the extraction of the associative concept within in-depth level of cognition was apprehensible (Fig. 4a & Fig. 4b). The reduction was performed by omitting <50% of ODC, independently to each groups (0.5x 0.051=0.025 for craftsmen and 0.5×0.076= 0.038 for designers). The highly-weighted associative words were identified as in-depth level of cognition with ODC score as follows (bold):
– Craftsmen (total: 202 words): 0.000 = 75words; 0.040 = 94 words; 0.054 =27 words; 0.067 = 6 words.
– Designers (total: 81 words): 0.000 = 48words; 0.083 = 23 words; 0.104 = 8words; 0.125 1 word; 0.146 = 1 word.
Hereafter, we select top 10 highly-weighted associative words from each group for further analysis.
Figure 4a Simplified concept networks of craftsmen’s in-depth level of cognition.
Table 2a Extracted verbal expressions of in-depth level of cognition (after reduction)
|Category||List of 202 extracted verbal expressions (ordered by the highest ODC score)|
|Craftsmen||body, clothes, idea, length, shape, waist, broad, butt, chest, corner, creativity, exercise, fresh, great, grow, grown, growth, ideal, impression, inch, oval, plaid, portion, replace, sample, size, slender, stripe, suggestion, tall, tight, weigh, wide, advice, bag, etc.|
Table 2b Extracted verbal expressions of in-depth level of cognition (after reduction)
|Category||List of 81 extracted verbal expressions (ordered by the highest ODC score)|
|Designers||fruit, produce, apple, basket, booth, orange, peel, silverware, style, tupperware, banana, carrot, cherry, common, computer, fig, grape, green, juice, lemon, market, olive, orchard, pear, position, process, public, rot, sale, soup, store, wait, etc.|
Up to this stage, the extraction of associative concept suggested that craftsmen tend to activate low associative concept. As it is demonstrated in high percentage of surface level of cognition 169/202 (83.6%). While designers activate more highly associative concept concerned to the issues that affect the presence of the fruit basket/container. As it is demonstrated significantly with a very high score of OCD and lower percentage of surface level of cognition 48/81 (59.2%). Following is identified characteristic of craftsmen’s and designers associative concept after reduction:
Table 3 Identified characteristic of craftsmen’s and designer’s associative concepts
|Category||List of identified characteristic|
|Craftsmen||(Proportion) length, inch, oval, portion, size, tall, tight, wide, centimeter, width, thin, thick, form, rectangle, measurement, narrow, weight, etc.
(Shape) body, shape, waist, butt, chest, corner, round, leg, hip, giant, cube, prism, etc.
(Operation) exercise, grow, replace, advice, bold, blend, bond, decision, firm, fit, perfect, stain, form, combine, cover, tie, trace, use, etc.
|Designers||(Companion) fruit, apple, orange, peel, banana, carrot, cherry, fig, grape, green, lemon, olive, orchard, pear, etc.
(Appealingness) salad, peel, juice, soup, process, produce, display, method, rotten, put, save, buy, shop, stand, fresh, etc.
(Scene) booth, silverware, tupperware, market, public, store, crowd, leaf, tree, wood, etc.
5.2 Analysis of Semantic Relation based on Factor Analysis
We distributed 40 associative words range from highest to low scores of ODC which correspond to previously identified characteristic of the associative concept (Table 3). The list of identified characteristics are proportion, shape, operation, companion, appealingness and scene, which determined to be 6 variables used in Factor Analysis. Furthermore, the correlation among variables was extracted into two factors with significant score of KMO 0,571. Factor matrix and corresponding name as follows:
Table 4 Rotated Factor Matrix
Table 5 Corresponding Name
Factor 1: Scene, Appealingness and Companion, hereafter referred to as Surroundings which pointed out issues that affect the presence of the fruit basket/container. Factor 2: Less Proportion, Less Shape, Operation, hereafter referred to as Object-Oriented which pointed out issues concerning technical aspects of of the fruit basket/container. Furthermore, factors were split into orthogonal map to invetigate semantic relation of the identified characteristic of craftsmen’s and designer’s associative concepts [Fig.5].
Figure 5 Semantic relation map
From the initial stage, the sorted verbal expressions showed that craftsmen paid attention on appearance and technical aspects of fruit basket/container. They described about: height, length, flat, stack, capacity, standard, coating, form of a duck, heron, etc. The extracted words obtained from concept network analysis was identified and its characteristic referred to proportion, shape and operation. In contrast, designers’ sorted verbal expression concerned on issues that affect the presence of the fruit basket/container, as they described about:
place, kitchen, hang, pluck, tree, wood, inform, remind, children, salad, dignity, etc. Some interesting comments were, “I don’t want to put it on table instead of hanging”, “Children is like being invited to reap fruit”, “How fruit container attractively served as salad for people”, “It is like a traditional banana-leaf container with more dignity”. The extracted words was identified and its characteristic referred to Companion, Appealingness and Scene.
We identify that craftsmen tend to activate low associative concept, as demonstrated in high percentage of surface level of cognition 169/202 (83.6%). While designers activate more highly associative concept as demonstrated by the high score of OCD and lower percentage of surface level of cognition 48/81 (59.2%). Refer to Associative Gradient Theory, the more closely associated or ‘stereotypical’ representations may lead to be less creative. the greater the number of association, the greater the probability of reaching a creative solution, because remotely associations (highly-weighted associative concept) are where creative solution occur (Mednick, 1962; Baer, 1993; Eysenck, 1997; Martindale, 1995). We also found approximately 24% of 202 extracted words derived from craftsmen were highly polysemous, and approximately 35% from designers. As Yamamoto et al., (2009) argue, the polysemy of a design idea has significant correlation with the originality of an idea. Therefore, it indicate that designers’ in-depth level cognition had greater probability of reaching creative solution .
The findings of this research suggested the role of closely and remotely associated concept at in-depth level of cognition at the early stage of idea generation differ how craftsmen and designers observe and define the design problem. Craftsmen’s in-depth level of cognition with less number of polysemous features explain their concerns on tangible-related issues such as proportion, shape, etc. While, designers’ in-depth level of cognition with more number of polysemous features concern on intangible-related issues such as surroundings-context (i.e. eating culture) and user’s affective preferences (i.e. companion, appealingness). The semantic relation map confirms craftsmen’s in-depth level of cognition is in position of concerning the physical property of the artifact, but less consider the surroundings and user’s affective preferences. Whereas designers much more concern on issues that affect the presence of the artifact and rather less attention to physical property.
In general, we can easily differ a designed artifact created by craftsmen and designers by describing the appearance. However, it is difficult to describe the nature creative cognition that differ their design thinking. This study have revealed the difference of in-depth level of cognition at the early stage of idea generation between craftsmen and designers. Further, these findings can be developed for a reference for co-creation educational program (design training) that suits to the craftsmen’s nature creative cognition, as they might develop their creative skills to produce more desirable products.
 Cross, N., “Designerly ways of knowing”, Birkhauser, 2006.
 Nagai, Y., Georgiev, G.V., Zhou, F., “A methodology to analyze in-depth impressions of design on the basis of concept networks”, Journal of Design Research, Vol. 9, No. 1, pp 44-64, 2011.
 Houtz, John C. and Patricola, Cathryn., “Imagery”, In Mark A. Runco; Steven R. Pritzker (Eds.), Encyclopedia of Creativity, Vol.2, 1999.
 Leijnen, S. and Gabora, L., “An agent-based simulation of the effectiveness of creative leadership”, Proceedings of the Annual Meeting of the Cognitive Science Society, August 11-14, Portland, Oregon. pp 5, 2010.
 Reingold, E. and Colleen, R., “Implicit cognition”, in Encyclopedia of Cognitive Science, Nature publishing group, 481–485, 2003.
 Georgiev, G.V. and Nagai, Y., “A Conceptual Network Analysis of User Impressions and Meanings of Product Materials in Design”, Materials and Design, Vol. 32, No. 8-9, pp 4230-42, 2011.
 T. Taura., E. Yamamoto., M. Y. N. Fasiha and Y. Nagai., “Virtual impression networks for capturing deep impressions”, Design Computing and Cognition DCC’10. J.S Gero (ed),. pp. xx-yy. © Springer, 2010.
 Ericsson, K. A. and Simon, H. A.,”Protocol Analysis: Verbal Reports as Data”, MIT Press, Cambridge, MA, 1993.
 Nelson, D. L., McEvoy, C.L., Schreiber, T.A., “The University of South Florida free association, rhyme, and word fragment norms”, Behav Res Methods, 36:402–7, 2004.
 Maki, W.S., Buchanan, E., “Latent Structure in Measures of Associative, Semantic and Thematic Knowledge”, Psychon Bull Rev, 15(3), 598–603, 2008.
 Batagelj, V., Mrvar, A., “Pajek – analysis and visualization of large networks”, In: Jünger M, Mutzel P, editors. Graph drawing software. Berlin: Springer, p. 77–103, 2003.
 Yamamoto, E., Mukai, F., M. Y. N. Fasiha., Taura, T and Nagai, Y., “A Method to Generate and Evaluate Creative Design Idea by Focusing on Associative Process”, Proceedings of the ASME2009. California, USA, 2009.