Design Creativity 2010 part 11 ppt

An Approach to Measuring Metaphoricity of Creative Design Hung-Hsiang Wang and Jung-Hsuan Chan National Taipei University of Technology, Taiwan Abstract. Metaphor is central to design creativity as it involves processes of discovering two different objects which are similar in some perspectives and combining them together into a new and meaningful one. This study argues the degree to which a design contains metaphor is a good indicator of the creativity of the design. In this paper the strength of metaphoricity is a function of feature similarity between its target and source entities, as well as the domain dissimilarity between the two entities. The situation of metaphoricity is the salience imbalance of the similar features between of its target and source entities. To test the argument, five award winners of various well-known creativity-oriented design competitions are accordingly presented to twenty-six design students to assess the metaphoricity strength and saturation, and the creativity on a subjective base. Results reveal the creativity has significantly positive relation between the object similarity and metaphoricity saturation. Keywords: design creativity, metaphor, similarity, industrial design 1 Introduction Metaphor is not only a style in speech and writing but a resourceful method of human’s thinking in daily life (Lakoff and Johnson ,1999). It has been thought the kernel ability of creativity that helps us to creatively put two things from two different domains together into a new one (Seitz, 1997; Ricoeur, 1981). Metaphors have long been recognized to play an important role in industrial design (Hey and Agogino, 2007). Metaphor is often used at earlier stages of conceptual design to solve problems or interpret meaning in a creative way. The conceptual design starting with an initial design goal, through ideation, evaluation, and finalization can be seen as a process of defining a target, searching sources to construct pairs of the target and sources as alternatives, and selecting a satisfactory one from these alternatives. Although there are some metaphor theories based on similarity measures, few of them have been applied to the area of design creativity. Moreover, empirical evidence supporting the positive relation between design metaphor and design creativity has rarely been reported. Therefore, this study aims to provide empirical evidence regarding design metaphor measures and its implications to design creativity. Quantitative results of questionnaires for assessing design metaphor and creativity are presented following a short literature review. 2 Metaphoricity in Design 2.1 Metaphorical Design Metaphors are represented by the form “A is B”, where B is called the source of the metaphor, and A is the target. Metaphor can be used for understanding of an unknown situation A in terms of one familiar thing B (Gentner 1983; Gentner and Markman, 1997; Novick, 1988; Vosniadou, 1989; Ortony, 1993). Interpretation of a metaphor is a process of discovering which features of the source may be valid and useful to understanding the target. To construct such a metaphor, one needs to find out the source B that is similar to the target A in some perspectives but dissimilar to each other in terms of membership of certain categories. The similarity maintains a reasonable mapping from the source B to the target A, while the dissimilarity promises an unusual mapping. The search for sources is thus described as a mapping of the target and sources based on their common features. As long as the mapping is reasonable but unusual to a certain degree, the conceptual design is said to be creative in terms of the processes or the products. A creative design is identical to a both new and meaningful design. Take Alessi's Anna G corkscrew, designed by Alessandro Mendini for example. The main goal of this project is to design a new object that belongs to the target domain of wing corkscrew. As it has the salient feature of two wing-like levers, it also called an angel corkscrew or butterfly corkscrew. Thus, dancers are selected as the source domain, and a female 90 H H. Wang and J H. Chan dancer’s body elements that are similar to the parts of a wing corkscrew are identified. On one hand, that the similarity between the wing corkscrew and an dancing woman make a reasonable mapping. The pairs of their similar features include (1) the handle of the corkscrew and the head of the angle, (2) the two levers and the two arms, (3) the rack and pinion connecting the levers to the body and the puff shoulder lace dress, (4) the motion in which the levers are raised as the worm is twisted into the cork, and the action of the angle’s raising arms while dancing, as well as (5) the smooth motion of pushing down the levers to draw the cork from the bottle and the elegant putting down arms. On the other hand, the similarity between human dancers to tools in household use is so low that the mapping is unusual. As a result, we can say Anna G corkscrew is a creative product because of good metaphor. Fig. 1. Anna G corkscrew (left) and the woman in puff shoulder lace dress (right). (adapted from http://www.alessi.com and http://www.costumediscounters.com/womens-costumes, respectively) 2.2 Metaphor and Design Creativity Metaphor is a very useful tool in creativity, not only in designing creative interface for effective and efficient use, but also in dreaming up both new and valuable ideas. Creativity enables designers to transcend conventional knowledge domain so as to investigate new ideas and concepts which may lead to creative solutions. As a metaphorical design is typically based on a reasonable and unusable mapping from source domain to target domain to represent some distinctiveness and meaningfulness, it has importance in design creativity. Statistically assessing the metaphors used by students in design creativity, Casakin (2006, 2007) determines synthesis of design solutions is the stronger factor of the use of metaphors, whereas metaphors play an important role in design creativity. Use of metaphors can contribute to designers’ (1) productivity of, meaningful, interpretable and relevant ideas, (2) rarity of the ideas, and (3) comprehensiveness of the ideas. These aspects are respectively associated to the three dimensions: fluency, originality, and elaboration used to assess divergent thinking and other problem-solving skills in Torrance Tests of Creative Thinking, developed by Torrance (1974). Furthermore, retrieving concepts from metaphors demands creative thinking. Effective and efficient indexing and retrieving the source objects that are similar to the target object, but belong to the domains that are dissimilar to the target domain are obviously related to fluency, flexibility, and originality in design creativity. Successful combination and adaptation of the features of the target and source objects are apparently associated with originality and elaboration. 3 Measuring Metaphoricity In this paper, the metaphoricity of a design is measured by the similarity between the target object and the source object, the dissimilarity between the target domain and the source domain, and the salience imbalance of the common features of the target object and the source object. The followings describe these three factors. 3.1 Object Similarity Similarity plays an important role in human perception (Goldstone, 1999; Kovecses, 2002; Tversky, 1977). Similarity measure used to quantify the degree of resemblance between a pair of cases (Liao, Zhang and Mount, 1998). There are many models of similarity measurement. The most common method in geometric (or spatial) models is an inverse measure of Euclidean distance. This method is suitable for continuous variables, though limited for discrete ones. However, similarity measures are commonly used for discrete features (Everitt et al., 2001). For real data sets, it is more common to see both continuous and discrete features at the same time. In other word, a database often contains such types of variables as binary, nominal, ordinal, interval, and ratio. A more powerful method is to use a weighted formula to combine their effects. A method for measuring mixed variables is proposed by Gower (1971) and extended by Kaufman and Rousseeuw (1990). The similarity measure for objects x and y with d features with mixed data (also called d-dimensional mixed data) is defined as   d i i d i ii 11 /)S(=y) S(x,  (1) An Approach to Measuring Metaphoricity of Creative Design 91 where Si indicates the similarity for the i-th feature (also called variable) between the two objects, and δi is Gower's General Similarity Coefficient. The coefficient δ i is usually 1 or 0 depending upon whether or not the comparison is valid for the i-th feature. If differential variable weights are specified, it is the weight of the i-th feature, or it is 0 if the comparison is not valid. That is, if the weight of any feature is zero, then the feature is effectively ignored for the calculation of proximities. Note that the effect of the denominator   d i i 1  is to divide the sum of the similarity scores by the number of variables; or if variable weights have been specified, by the sum of their weights. Calculation of the component similarity S i is various with discrete and continuous variables. For the discrete variables (including binary), S i is assigned to either 1 if x i = y i , or 0 if x i ≠ y i . For the continuous variables, S i is obtained by using the normalized city- block distance as S i = 1 - |x i - y i | / R i (2) where R i is the range of the i-th feature over the two objects Again, take Anna G for example. In the two- dimensional mixed data as shown in Table 1, the target is Anna G corkscrew, denoted by x, and the source is the female dancer, denoted by y. Table 1. Mixed variables for the target and source objects of Anna G Feature Discrete (structural) Continuous (behavioral) i-th 1 2 3 4 5 Target object x yes yes yes 0.9 1.0 Source object y yes yes yes 0.8 0.7 Coefficient δ i 1 1 1 2 2 Similarity S i 1 1 1 0.9 0.7 Note: Each i-th feature denotes as the followings. 1: a head-like part attached to the top of body, 2: two arms-like parts attached to shoulder, 3: puff-shoulder-like shape on each shoulder, 4: rotating the head-like part while raising two arms, 5: smooth pushing down two arms-like parts two arms. For simplicity, let’s assume that behavioral features such as rotating are twice as important as structural features such as having arms. Thus, the weights of the former are given by 2, while that of the latter 1. Furthermore, the behavioral and structural features are treated as continuous and discrete features, respectively. The range of each continuous feature is given by 1. Thus, the similarity measurement is obtained as S (x, y)=(1×1+1×1+1×1+2×0.9+2×0.7) / (1+1+1+2+2) = 0.89 3.2 Domain Dissimilarity In addition to the similarity between the target and source objects, the dissimilarity between the target and source domains also plays an important role in metaphorical design. Winner (1985) suggests a good metaphor have a sufficiently long distance (i.e., higher dissimilarity) between the domains to which the target and source objects correspondingly belong. Casakin (2005) points out that the degree of difficulty to establish a metaphor is mainly determined by how remote the source is from the target. Michalko (2001) also determines a positive relationship between the probability of inspiring new concepts by metaphors and the domain dissimilarity. This study measures the distance between the two classes or categories of which the target and source objects are members, respectively, to obtain the domain dissimilarity. For the target, Industrial and Business Taxonomy, developed by Ministry of Economic Affairs of Taiwan, is a practical domain classification. For example, the classes can be home accessories, 3C-electronics, transportation, fashion, and sport and entertainment. In contrast, the source domains are much more diverse. They may range from nature to artificial, from creature to non-creature, or from tangible to intangible classes. The domain taxonomy seems to be arranged in a hierarchical structure, which is typically organized by supertype-subtype relationships. In such an inheritance relationship, the subtype by definition has the same features as the supertype plus one or more additional features. For example, corkscrew is a subtype of wine accessory, but not every wine accessory is a corkscrew. Hence, a type must satisfy more features to be a subtype than to be a supertype. Theoretically the domain dissimilarity can be computed not only by the inversed similarity, but also by the depth and width of the supertype-subtype relationships. Sometimes it is hardly to consider such relationships because of the difficulty of specifying the consistent supertype of the target and source objects. For instance, the supertype of wing corkscrew could 92 H H. Wang and J H. Chan be corkscrew, wine accessory, tool, or to the extreme, thing. Likewise, the female dancer could be the subtype of female, human being, mammal, animal, or to the extreme, thing, too. At this moment, it is more or less uncertain to decide which hierarchical level of supertypes. However, given that the supertypes of target and source objects have not recognized yet, we can judge the domain dissimilarity by simply estimating the distance between the undecided supertypes without naming them or specifying their detailed features. For example, we can assess this domain dissimilarity by giving a value, 0.9, for the dissimilarity between the category of wing corkscrew and the category of female dancer. 3.3 Salience Imbalance Besides similarity of objects and dissimilarity of domains in metaphors, the salience (i.e., significance) of the common features between the target and source objects plays an important role. On the basis of Tversky’s (1977) notion, Ortony (1979) thinks the imbalance, denoted by I(x, y), in salience levels of matching features of the two objects is a principal source of metaphoricity. Given that the feature sets of the target object x and the source object y are A and B, respectively. The salience imbalance of x and y, denoted by I(x, y), is expressed as a linear function of the measures of their common features, and is given by I(x, y) = g(ƒ A (A∩B)–ƒ B (A∩B)) (3) where (A∩B) represents the of common features of x and y, ƒ A and ƒ B represent measures of salience based on the values in A and B respectively, and g is some, probably additive, function. Ortony (1979) suggests that a convenient way of conceptualizing this imbalance is to visualize the features of x and y as a list with the most salient features at the top. Then salience imbalance can be thought of as the degree of slope from features in B to features in A, and can be characterized, to a first approximation, by considering the combined effect of the difference in salience between the matching features for x and for y together with the (independent) degree of salience in each, as in Equation (3). Using the concept of salience imbalance, Ortony et al. (1985) classify four types of similarity into literal similarity, metaphorical similarity (including simile), anomalous similarity, and reversed metaphorical similarity. If the common feature salience is both high in the target and source objects, the similarity is literal. For example, the two objects may be almost identical, or one of the objects is obviously the explanation of the other. On the contrary, if it is both low in the target and source objects, similarity is anomalous because such a resemblance is too trivial. If the salience is high in the source object, but low in the target object, the similarity is metaphorical. In contrast, if the salience is low in the source object, but high in the target object, it is called reversed metaphorical similarity. This classification can be represented in diagonal arrow lines from the salience ranking of source features to that of target features as shown in Table 2, developed by Wang and Liao (2009). This diagram of salience imbalance analysis allow us to (1) list as many features of the target object and the source object in salient order, respectively, (2) link the pair of two similar features by drawing an arrow line from the source feature to the similar target feature, and (3) assign the degree of similarity between the two objects on the linking lines. As the slope of these linking lines describes the degree of metaphorical similarity, this diagram is a useful tool of questionnaires for the subjects to depict their responses about metaphoricity. Table 2. Diagram for salience imbalance (adapted from Wang and Liao, 2009) For representing the difference between the target and source objects, an exaggerative but reasonable way to deal with the salience ranking is required. This study considers the law of diminishing marginal utility to convert the salience ranking into a non-linear decreasing sequence as salience weighting. There are many popular decreasing sequences, such as 1/n, 1/2 n- 1 , and n 2 (n=1, 2, 3,…), used for ranking transform. Wang and Chou (2010) compare the exaggerative effects of the three sequences and conclude that the decreasing sequence, 1/n, is superior to the others. For the object x with d features, the i-th feature’s normalized salience is given as    d i xi iiw 1 )/1(/)/1( (4) Target object x Source object y Features of x Salience in x Salience in y Features of y x i Higher w i Higher w i y j x n Lower w n Lower w n y n Literal Anomaly Metaphor Reversed Meta p hor An Approach to Measuring Metaphoricity of Creative Design 93 For example, the sequence of ranking salience, 1, 1/2, 1/3, 1/4, 1/5, is normalized into the sequence of rating salience 0.438, 0.219, 0.146, 0.109, 0.088. Furthermore, Wang and Chou (2010) propose a practical way to determine the feature salience imbalance of the target object x and the source object y as   d 1i d 1i xiyiixiyi )w-(w / )Sw-(w= y) I(x, (5) where S i is the similarity of the i-th feature of the target object x and the source object y. S i can be obtained, as the equation (2), but not limited to this method. By adding two features to the data in Table 1, let create Table 3 for demonstrating how to calculate I(x, y). Given that we consider only features of the target and source objects, in which only five features are similar (S i >0), and the rest are absolutely dissimilar (S i =0). The normalized salience values converted from salience rankings are shown in Table 3. Thus, the salience imbalance of the objects x and y is computed as I(x, y) = (0.257×1+0.097×1+0.052×1+0.032×0.9+0.022×0.7) / 0.46 = 0.450/ 0.46 = 0.978 > 0 Table 3. Mixed variables for the target and source objects of Anna G Feature i-th 1 2 3 4 5 6 7 Similarity S i 1 1 1 0.9 0.7 0 0 Rank. Salience in x 3 4 5 6 7 1 2 y 1 2 3 4 5 6 7 Norm. Salience: w x .129 .096 .077 .064 .055 .386 .193 w y .386 .193 .129 .096 .077 .064 .055 w y –w x .257 .097 .052 .032 .022 (w y –w x ) .46 3.4 Metaphoricity As previously described, the salience imbalance, I(x, y), is practical for identifying whether or not an object is a member of metaphorical design. For a typical metaphorical design, its salience imbalance value is supposed to be as greater than 0 as possible. Also, the metaphoricity strength, T(x, y), of a design can be thought as a function of the feature similarity and domain dissimilarity between the target and the source. This study defines it as T(x, y)= (α×S(x, y)+β×D(x, y))/ (α+β) (6) where α, β are the weights for the feature similarity and domain dissimilarity, respectively (α+β≠0) By the definition, the design example shown in Tables 1 and 3 is a significantly typical metaphor. This study calls this characteristic “saturation”. The metaphoricity is extremely saturated, because the salience imbalance, I(x, y), is 0.978. Moreover, this well-saturated metaphorical design is significantly of strength, for T(x, y), is 0.895 (= (0.89+ 0.9)/ 2), given α= β= 1. 4 Example and Testing To determine the relation between the metaphoricity and creativity of designs, this study chooses five metaphorical products as the stimuli for testing, in a fashion of purposive sampling. The stimuli are chosen from five international competitions: International Forum (iF) concept award; red dot Design Award- concept; Good Design Award (G-Mark); International Design Excellence Award (IDEA) and Taiwan International Design Competition- students (TID), as displayed in Table 4. Participants of this test are twenty-six industrial design students of National Taipei University of Technology. Table 4. Five stimuli for testing Title Image Target Source Award Aroma Humidifier humidifier potted plant G-Mark Jellyclick laptop mouse jelly IDEA Orangin juicer pencil sharpener red dot Pebble Eraser eraser pebble TID Zipper Speaker speaker zipper iF 94 H H. Wang and J H. Chan First, the materials shown in Table 4 are presented to each participant. He or she is requested to complete the following stages for each design: 1. List the top-seven salient features of target and source respectively. 2. Specify the salience rankings for the seven target features and the seven source features respectively. 3. Link up the pairs of similar features by drawing an arrow line from the source feature to the similar target feature. 4. Put the degree of the similarity on each line (ranging from 0 to1). 5. Determine the degree of the dissimilarity between the target domain and source domain (ranging from 0 to1). 6. Determine the degree of overall creativity of this stimuli (ranging from 0 to1). The metaphoricity of all the five stimuli is measured by using the raw data acquired in the above stages. For convenient reason, the strength constants α and β, and the coefficient δ i for each feature are set as 1. As the space is limited, let’s merely take one of the participant’s responses on Zipper Speaker for example. Table 5 shows how the diagram of salience imbalance analysis is applied. This participant identifies the top-seven features of the speaker, but has some difficulty on the sixth and seventh features of the zipper. Although only top-five features of the zipper are listed, the normalized salience used is still based on seven features without any difficulty, for it is impossible to have a pair including the sixth or the seventh features. The participant then draws arrow lines to connect the common features, and put the similarity value of each pair of common feature on the corresponding line. The similarity between a speaker and a zipper is thus obtained as S(speaker, zipper)= (0.4+0.8+0.5+0.7) / 4 = 0.6 Since the participant gives the degree of the dissimilarity between the target domain and source domain, D(speaker, zipper), as 0.8. Consequently, the metaphoricity strength is obtained as T(speaker, zipper) = (S(speaker, zipper)+D(speaker, zipper))/ 2= (0.6+0.8)/ 2= 0.7 The summation of salience imbalance differences is computed as (w y –w x )= (0.386–0.129)+(0.193–0.096)+(0.096– 0.055)+(0.077–0.193)= 0.279 The feature salience imbalance of the Zipper Speaker is then calculated as I(speaker, zipper)= ((0.386–0.129)×0.4+(0.193–0.096) ×0.8+(0.096–0.055)×0.7+ (0.077–0.193)×0.5) / 0.279= 0.54 > 0 Table 6 presents results of measuring object similarity, domain dissimilarity, metaphoricity strength, salience imbalance( metaphoricity saturation), and creativity for each stimulus. In general, the relation between the metaphoricity strength and the creativity is intermediately positive (r=0.65). Nevertheless, the correlation coefficient of the domain dissimilarity and the creativity is rather low (r=0.08), whereas the correlation coefficient of the object similarity and the creativity is significantly high (r=0.90). The implications are two-fold. First, this relation between is by no means a perfect linear correlation, if the measurement of domain dissimilarity is applicable. The metaphoricity strength becomes much more undecided than this study predicts. Having got this point firmly recognized, in our short study the weight of the feature similarity, α, and the weight of the domain dissimilarity β should not be set to 1. Alternatively, we can turn to only consider the object similarity instead of the metaphoricity strength. Table 5. Diagram of salience imbalance analysis for Zipper Speaker Target: Speaker Source: Zipper Features (Normalized) Salience (Normalized) Salience Features Broadcast music (0.386) 1 1 Control opening/closure Rotating-button (0.193) 2 2 Moving up and down Control volume (0.129) 3 3 Jagged parts On/off (0.096) 4 4 Two in one Square box (0.077) 5 5 Pull ring Connect. comp. (0.064) 6 6 - Couple (0.055) 7 7 - 0.4 0.8 0.5 0.7 An Approach to Measuring Metaphoricity of Creative Design 95 Second, it might be too abstract for the participants to learn what the target domain and source domain of an object are. Perhaps, determining the supertype of a subtype, or the class of an object, is not as straightforward as determining the features of the subtype, or the features of an object. In the test, a few participants ask for clear definition or exemplars, when they are requested to describe these two domains for each stimulus. The above two points remain to be proved in further investigations. Table 6. Results of metaphoricity and creativity measurements Title Object Similarity Domain Dissimilar Metapho -ricity Strength Metapho- ricity Saturation Creativity Aroma Humidifier 0.78 0.46 0.48 0.45 0.75 Jellyclick 0.75 0.32 0.55 0.78 0.79 Orangin 0.62 0.32 0.41 0.22 0.61 Pebble Eraser 0.54 0.36 0.46 0.26 0.55 Zipper Speaker 0.71 0.46 0.45 0.31 0.62 In contrast, the metaphoricity saturation (i.e., salience imbalance) and the creativity have considerable positive relation (r=0.88). The correlation coefficient is as high as that of the object similarity and the creativity (r=0.90). This represents that the two factors can be used as an alternative indicator of creativity of designs. 5 Conclusion This research has proposed a feature-based approach to measuring metaphoricity of designs, including measures of the object similarity, domain dissimilarity, and salience imbalance. The strength of metaphoricity is defined as a function of feature similarity between its target and source entities, as well as the domain dissimilarity between the two entities. The saturation of metaphoricity is the salience imbalance of the similar features between of its target and source entities. To test the argument, five award winners of various well-known creativity-oriented design competitions are accordingly presented to twenty-six design students to assess the metaphoricity strength and saturation, and the creativity on a subjective base. Results reveal the creativity has significantly positive relation between the object similarity and metaphoricity saturation. In this sense, creative designers are those who learn how to maximize the similarity between the target and source objects, the dissimilarity between the target and source domains, and the salience imbalance, in order to create both new and meaningful solutions. Nonetheless, relation between the creativity and the domain dissimilarity might not be a perfect linear correlation, which is much more uncertain than predicted. The strength of metaphoricity remains to be determined in further studies. The limitation of this method is that features of target and source may hard to indicate by general participants with non-design background, and participants with different culture would evaluate metaphorical design differently. To sum up, metaphoricity measures have potential to develop alternative tool for assessing the creativity of designs. Acknowledgements We wish to thank National Science Council, Taiwan, ROC, for their generous financial assistance under Grant NSC 99-2221-E027 -084. References Casakin HP, (2005) Design aided by visual displays: a cognitive approach. Journal of Architectural and Planning Research 22(3):250–265 Casakin HP, (2006) Assessing the use of metaphors in the design process. Environment and Planning B: Planning and Design 33(2):253–268 Casakin HP, (2007) Metaphors in design problem solving: implications for creativity. International Journal of Design 1(2):21–33 Everitt B, Landau S, Leese M, (2001) Cluster analysis. 4 th edition. London: Arnold Gentner D, Markman AB, (1997) Structure mapping in analogy and similarity. American Psychologist 52(1):45– 56 Gentner D, (1983) Structure-mapping: a theoretical framework for analogy. Cognitive Science 7:155–170 Goldstone RL, (1999) Similarity. In Wilson RA, Keil FC, (Eds.), The MIT encyclopedia of the cognitive sciences, 757–759, Cambridge, MA: MIT Press Gower J, (1971) A general coefficient of similarity and some of its properties. Biometrics 27:857–872 Hey HGJ, Agogino MA, (2007) Metaphors in conceptual design. 2007 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference Kaufman L, Rousseeuw P, (1990) Finding group in data: an introduction to cluster analysis. New York, NY: John Wiley & Sons 96 H H. Wang and J H. Chan Kovecses Z, (2002) Metaphor: a practical introduction. NY: Oxford Lakoff G, Johnson M, (1999) Philosophy in the flesh: the embodied mind and its challenge to western thought. New York: Basic Books Liao TW, Zhang Z, Mount C, (1998) Similarity measures for retrieval in case-based reasoning systems. Applied Artificial Intelligence 12(4):267–288 Michalko M, (2001) Cracking creativity: the secrets of creative genius. California: Ten Speed Press Novick LR, (1988) Analogical transfer, problem similarity, and expertise. Journal of Experimental Psychology: Learning, Memory, and Cognition 14(3):510–520 Ortony A, (1979) Beyond literal similarity. Psychological Review 86(3):161–180 Ortony A, (1993) Metaphor and thought: the role similarity in similes and metaphors. Cambridge University Press, New York, 342–356 Ortony A, Vondruska JR, Foss AM, Jones EL, (1985) Salience, similes, and the asymmetry of similarity. Journal of Memory and Language 24:569–594 Ricoeur P, (1981) Metaphor and the central problem of hermeneutics. In Thompson JB, (Ed. and Trans.), Hermeneutics and the human sciences, 165–181, Cambridge, England: Cambridge University Press Seitz JA, (1997) The development of metaphoric understanding: implications for a theory of creativity. Creativity Research Journal 10(4):347–353 Torrance EP, (1974) Torrance tests of creative thinking: directions manual and scoring guide (figural test, form A) (revision). Princeton, N J: Personnel Press Tversky A, (1977) Features of similarity. Psychological Review 84(4):327–352 Vosniadou S, (1989) Analogical reasoning as a mechanism in knowledge acquisition: a developmental perspective. In Vosniadou S, Ortony A, (eds) Similarity and analogical reasoning, 413–437, Cambridge University Press, Cambridge Winner E, (1985) Invented worlds: the psychology of the arts. Baker & Taylor Books (Chinese translated by Tao DF, 1997), Taipei: Gardencity Wang HH, Liao WJ, (2009) Applications of metaphor theory to product design. In Proceedings of International Association of Societies of Design Research 2009 (CD- ROM), Seoul, Korea Wang HH, Chou CP, (2010) Wind, water, and stone: a cultural product design of applying design metaphor. In Proceedings of 2010 International Conference of Innovation and Design, Taipei, Taiwan, 64–69 Interrelations between Motivation, Creativity and Emotions in Design Thinking Processes – An Empirical Study Based on Regulatory Focus Theory Madeleine Kröper 1,2 , Doris Fay 2 , Tilmann Lindberg 1 and Christoph Meinel 1 1 Hasso Plattner Institute at University of Potsdam, Germany 2 University of Potsdam, Germany Abstract. Design thinking, here defined as a team-based innovation method, helps to deal with complex design problems by sustaining in-depth learning processes on problem perception and diverse solution paths. To carry out design thinking processes successfully, motivation is a central psychological aspect to ensure creativity of the project outcome. In this paper, we ask how motivation is affected by the design thinking process and how it is related to team member’s emotions throughout the process. We adopted regulatory focus theory to conceptualize motivational variables. Experience Sampling Method within a field study with two samples was used, investigating people’s motivation of setting and approaching goals throughout real-life design projects that used design thinking. Results of this study show that the different phases carried out in design thinking processes significantly impact motivation and emotions of the members of a design team. Keywords: Design Thinking, Design Thinking Processes, Motivation, Creativity, Emotions, Teams, Regulatory Focus Theory 1 Introduction In the broadest sense, design thinking refers to the “study of cognitive processes that are manifested in design action” (Cross, Dorst and Roozenburg, 1992). Practitioners as well as scholars in various disciplines have long been interested in understanding the cognitive processes that underlie design activities. Early research trying to unravel the thought processes in design activities studied how outstanding designers approach problems and develop creative solution concepts (e.g. Lawson, 2006; Cross, 2007). This research has initiated an extensive scientific discourse on the exploration and analysis of cognitive strategies that carry the generation, synthesis and creative transformation of divergent knowledge within design processes (e.g. Nagai and Noguchi, 2003; Owen, 2007). Identified design strategies have been reinterpreted as normative guidelines for design projects and creative problem solving in general (Lindberg, Noweski and Meinel, 2010). In this context, design thinking has been translated into a holistic framework moving beyond designers’ professional domains and it has since been gradually applied to various disciplines and fields of innovation in both academia and business (Beckman and Barry, 2007; Brown, 2008; Dunne and Martin, 2006). The fundamental principle underlying design thinking is that design problems and solutions are explored in parallel in consideration of different stakeholder perspectives (Cross, 2007; Lawson, 2006). Design problems are regarded as made up of exogenous stakeholder perspectives (the user’s, the client’s, the engineer’s, the manufacturer’s, the law- maker’s, etc.) that finally decide about the solution’s viability (Dorst, 2006). Dealing with a design problem’s complexity is therefore a matter of negotiation between different and probably conflicting perspectives, so that design processes are regarded as a “reflective conversation with the situation” (Schön, 1983). Design thinking thus supports all activities relevant for accessing the diverse knowledge and multiple perspectives that reside in the different stakeholders in order to use them for inspiration; and it facilitates the creative transformation of the knowledge base into new concepts. The specific problem solving patterns in design thinking are rather determined by heuristic and situational reasoning than by analytical and rationalist thinking. Furthermore, instead of external standards for evaluating the quality of design outcomes, design thinking asks for developing those standards within the process. Therefore, design thinking assigns strong responsibility for deciding and evaluating how to proceed in a design process to the design team itself (that is what knowledge should be grasped and what concepts and designs should be elaborated). As a result, design thinking process models cannot be more 98 M. Kröper, D. Fay, T. Lindberg and C. Meinel than a framework of suggestions that help design teams to go through their own learning and creativity processes. Against that background, we assume that team motivation plays a decisive role in putting those suggestions into practice. We therefore seek to find out how motivation is affected by the different phases of the design thinking processes; this will enable us to better understand team creativity. We also explore whether motivation and emotions in design thinking processes are interrelated, as both concepts show strong interdependencies (Ryan, 2007). To deal with these questions, we draw upon a conceptionalization of motivation offered by regulatory focus theory (Higgins, 1997; 1998). We conducted a study using the Experience Sampling Method with design teams. Design teams adopted design thinking methodology; they worked in two German IT companies. In the following, we present the conceptual and theoretical foundations and develop this study’s hypotheses. 1.1 Design Thinking Process Model This study draws on a comprehensive design thinking process model that has been formalized at the Hasso- Plattner-School of Design at Stanford (US) and the HPI School of Design Thinking in Potsdam (Germany). It distinguishes six phases (Plattner, Meinel and Weinberg, 2009): understand, in which a design team is asked to build up general expertise about a design problem, to identify stakeholders and contexts of usage for further examination; observe, in which the design team goes into the field and gathers widespread insights and develops empathy for the stakeholders of the design problem; synthesis/point of view, in which the collected insights are summarized, shared in the team, and compiled in a framework of viewpoints on the design problem; ideate, in which – based on the lessons learned so far – ideas and concepts are created (for instance by brainstorming techniques) and roughly sketched out; prototyping, in which ideas and concepts are turned in tangible representations allowing to generate genuine feedback from users and other stakeholders; and test, in which this feedback is collected and processed for further refinements and revisions. As Figure 1 shows, these phases are not placed in a linear sequence, but are highly iterative. Therefore, the responsibility for the decision on when to move into which phase and how to get through an entire design process lies with the design team. The model is complemented by a set of rules that communicates a certain mind-set towards creative design. Rules emphasize 1) the readiness to explore seemingly odd paths as well (instead of going rashly for the obvious things) and 2) acting generally quickly, experimentally, and iteratively. Those rules are in particular: “fail often and early”; “defer judgement” and “encourage wild ideas” (cf. Osborn, 1953). Fig. 1. Iterative design thinking process (Plattner, Meinel and Weinberg ,2009) 1.2 Regulatory Focus Theory and Creative Performance We draw on regulatory focus theory to explore motivation in design thinking (Higgins, 1997; 1998). This theory presupposes that human motivation serves to satisfy the two basic needs of approaching pleasure and avoiding pain (hedonic principle). The theory suggests that these desired hedonic end-states are reached through self-regulatory processes, which refer to the processes by which people seek to align themselves with appropriate goals or standards (Crowe and Higgins, 1997). Two distinct types of regulatory systems, called promotion and prevention focus, drive this process of self-regulation. The promotion focus has a desired end-state as reference value, focusing individuals on goals they long for and is induced by nurturance needs, ideals and rewards (gain/no-gain situations). The prevention focus, conversely, has an undesired end-state as reference value, motivating individuals to avoid damages or unpleasant situations. This focus is induced by security needs, duties and the fear of punishment (non-loss/ loss situations). It is assumed that the promotion focus represents the “ideal self“, that is a person’s wishes, hopes, and aspirations, while the prevention focus represents the “ought self“, which includes a person’s obligations, duties, and responsibilities (Higgins, 1997). Both foci influence people’s perception, behavior, performance, and emotions (Förster and Higgins, 2005). The theory distinguishes furthermore between chronic and momentary foci. Individuals differ in their chronic tendency to be promotion and prevention oriented; furthermore, signals and stimuli of any type of situation also activate the promotion and/or prevention focus (Higgins, 1998; Crowe and Higgins, 1997). Thus, process feedback, task instructions or goal framing has a significant impact on the two dimensions of regulatory focus (Idson, Liberman, and Higgins, 2004; Higgins, Shah, and Friedman, 1997). . them have been applied to the area of design creativity. Moreover, empirical evidence supporting the positive relation between design metaphor and design creativity has rarely been reported empirical evidence regarding design metaphor measures and its implications to design creativity. Quantitative results of questionnaires for assessing design metaphor and creativity are presented. meaningfulness, it has importance in design creativity. Statistically assessing the metaphors used by students in design creativity, Casakin (2006, 2007) determines synthesis of design solutions is the