You are on page 1of 26

DEVELOPING A METHODOLOGY FOR DESIGN FOR AESTHETICS BASED ON ANALOGY OF COMMUNICATION

Ernest J. J. van Breemen Willem G. Knoop Imre Horvth Joris S. M. Vergeest

Delft University of Technology Subfaculty of Industrial Design Engineering Jaffalaan 9, NL-2628 BX Delft, The Netherlands Phone: +31 15 278 3520 Fax: +31 15 278 1839 E-mail: i.horvath@io.tudelft.nl Binh Pham School of Information Technology & Mathematical Sciences University of Ballarat PO Box 663 Ballarat, VIC 3353, Australia Phone +61 3 5327 9286 Fax +61 3 5327 9289 Email: b.pham@balarat.edu.au

Abstract
Although attempts have already been made, computer support of industrial design is still in its infancy, especially of design for aesthetics. The reason is that no methodology is available to incorporate in computer support, design aspects like appearance, pleasantness and human usage of a product. The objective of the paper is to present a methodology that facilitates development of computer support of design for aesthetics. While a designed product can trigger definite aesthetic responses to observers, it is not easy to relate these responses to the characteristics of the product. For that reason the paper focuses on the issues related to a practical coupling of intended aesthetic impressions and shape design. First it summarizes the fundamentals and knowledge related to aesthetics. Then it compares two studies on phenomenological and systematic approaches that have been carried out to model user responses to products and to identify possibilities for designers to influence those responses with the geometric design of products. Based on the generalization of the experiences a novel methodology for unifying aesthetics and design is introduced following the analogy of information communication. The problem of realizing aesthetic intents is decomposed into an activity loop that extends to statistical, syntactical, semantical and pragmatic levels of communication. Our initial experiments and tests have shown that the information communication analogy helps understanding the core problem of design for aesthetic and supports the elaboration of a quasi-formalized methodology that serves well as a base of computer tool development.

1 Introduction
The roots of understanding the aesthetics of natural and artificial things trace back to the distant past. Some two thousand years ago, the roman architect and artist Vitruvius claimed that, for human products, firmitas, utilitas and venustas are the key ingredients, as well as their harmony. Henry van de Velde, a Belgian architect and designer, said Beauty is the result of clarity and system, and not an optical illusion (De Wolff, 1992). Mijksenaar (1997) developed a scheme to measure the relative strength of these qualities for industrial design products. Nevertheless, the most fundamental interrelations have not been fully explored and clarified yet. There are several associations to explain, e.g., how the aesthetic quality is experienced: based on unconscious visual perception or, on the contrary, based on rigorous analysis and comparison with already acquired patters; what is the system of qualities we are looking for in an aesthetically pleasing product; what are the rules of formation of aesthetics; or can the qualities delivered by aesthetics be matched against any set of formalized expectations. In pursuit of developing increasingly complex products for a global market where the competitive products deliver almost the same functionality and where completely different aesthetic expectations are to be fulfilled, design for aesthetics gets more and more emphasis. With balanced prices, aesthetics is often the only factor that makes a difference and results in success or failure. Nevertheless, design for aesthetics is in lack of the proper understanding of the theoretical fundamentals and the needed computer support. Naturally, designers want to incorporate aesthetic intents into their products, but there are several problems they have to deal with. First, we have to mention the diversity of possible responses of customers to a single product. Even if the targeted groups of users are specified and their expectations are circumscribed, it is still an open issue how to translate the aesthetic properties that evoke the requested response to a product. One source of vagueness is that the language to describe aesthetics is very rich, diverse and fuzzy, where one term may have a number of meanings to different people, or many terms may imply the same or very similar meaning. Another reason is that many different characteristics of the product may act singly or in combination to evoke the responses. Furthermore, the responses may depend on other factors besides the product itself such as previous experiences or cultural background of the observers. The latter becomes more clear when we consider the aesthetic response as a feeling based response. According to Frijda (1986) feelings influence actions and that is what happens when we respond to an aesthetic stimulus for instance an action to buy a product. However, feelings are very personal so it can be explained that personal experiences and cultural background have their impact on the aesthetic response. We thus need to identify the characteristics of aesthetics and products that are relevant to the task of design for aesthetics. Our research into the state of including design for aesthetics into product design and development explored different approaches. The most elementary, and most widespread in the industry, relies completely on the aesthetic education and intention of the product designers. Other approaches are based on various levels of understanding the relationship between aesthetic and design. The methods that are focusing on one or

several aspects of design for aesthetics, but do not strive after being comprehensive or exhaustive will be referred to as phenomenological methods. Those however, that intend to cover all aspects and to put together all means in order to be able to manipulate a shape for the sake of achieving a particular impression and/or feeling will be called systematic methods. The authors are going to present details about how these methods are applied in the practice and how much they support design for aesthetics. Computer-aided conceptual design and shape design is expected to support design aspects other than covered by conventional Design for X techniques. Aesthetic design, together with design for ergonomics, user friendliness, adaptability, etc. is one of the fields which will surely be in the focus of the research and development in the near future due to its role in enhancing product acceptance. These aspects however fall almost completely out of the scope of the recently existing CAD technology. In the background of the observed situation are the following facts: (a) the explicit request for computer support in the field of design for aesthetics emerged only a couple of years ago, (b) relationship between shape and aesthetics as well as between geometric and psychological aspects is not fully understood neither from theoretical nor methodological point of view. To provide effective computer support to this field of conceptual and shape design, on the one hand, proper understanding of aesthetics is needed and, on the other hand, a computer-orientated design for aesthetic methodology is required. In this paper the authors explore the factors that influence the reasoning on and process of design for aesthetics. Based on their initial investigations they present a framework and a methodology that help develop of proper computer support. Chapter 2 starts with identifying the characteristics of aesthetics and products that are relevant to the task of design for aesthetics. This chapter also presents some findings from past research and distinguishes two ways (phenomenological and systematic) of coping with the relation between aesthetics and shape. These two approaches are respectively elaborated in chapters 3 and 4. Our experiences lead to the proposition of a methodology for mapping aesthetics and shape in chapter 5. The theory of the Jacobian matrix is used to create a mathematical foundation. Finally, in chapter 6, we will discuss how our proposed methodology can be applied to the development of computer support for aesthetic design.

2 Background on aesthetics and shape


2.1 Aesthetic characteristics The term aesthetics is broadly used to describe the characteristics of the appearance of a design. In particular, it refers to the responses that indicate the degree of discrimination in perception when people are confronted with the design. This perception depends on individual interpretation which may arise from emotional responses or comparison with previous experience. Ruskin (1971) discussed the contrasts between the discriminatory manner of this type of perception (e.g. by artists) and the cursory manner of normal perception. The latter type is more utilitarian and ignores information that is not essential for everyday life. The concept of aesthetics has been extensively analyzed by philosophers for the last three centuries, e.g. Kant, (1790), Tolstoy, (1960), Bearsdley, (1958). Early theory of taste in the eighteenth century sees aesthetics as reactions in an observer, that are

triggered by a specific kind of object. Thus, five essential components are involved: faculty of perception, faculty of reaction, an object to be perceived, a mental state resulted from reaction to the object and a judgment of taste. On the other hand, the attitude theory believes that the appreciation of aesthetics is more subjective and requires certain modes of perception or consciousness from the observer (Dickie, 1974). The concepts of style and originality are also often connected with aesthetics. Chen and Owen (1997), propose a method to systematically describe style profiles of products. A style refers to designs which possess a number of recognizable common characteristics (e.g. Art Deco), while taste refers to personal preferences, sensitivity or appreciation of certain type of beauty or style. Although the originality of style gives rise to the singular individuality of a design and often enhances its value, aesthetics does not necessarily imply originality or vice versa. Beardsley (1958) who viewed aesthetics as a philosophy of criticism, gave a thorough analysis of aesthetic objects in literature, visual arts and music, and discussed the nature of critical evaluation of aesthetics. Since then, many philosophers have attempted to formalize the properties and meanings of aesthetics for evaluative purposes. Goldman (1995) proposed a classification of evaluative aesthetic terms into the following eight categories: Broadly evaluative, e.g. beautiful, ugly, sublime, dreary. Formal, e.g. balanced, graceful, concise. Emotional, e.g. sad, angry, joyful, serene. Evocative, e.g. powerful, stirring, amusing, hilarious, boring. Behavioral, e.g. sluggish, bouncy, jaunty. Representational, e.g. realistic, distorted, artificial. Perceptual, e.g. vivid, dull, flashy. Historical, e.g. derivative, original, conservative. These terms which have been used for art criticism and evaluation, describe clearly the typical reactions of an observer to an object. What seems to be elusive is the relationship between these terms and the characteristics of the object in question. It has been commonly accepted by many researchers in this field of philosophy that there are three basic characteristics of the work (like product characteristics in design) that could be used for evaluation: expression, representation and form. An art work (or a design) is expressive if it arises some emotion from an observer. This emotion may be embedded in the work by an artist (or a designer) on purpose or unintentionally. Representation refers to the content of the design art work (design), which may be actual, idealized or imagined, while form refers to the totality of the shape and the structure, organization and composition of an object. Representation and form are the most easily identifiable properties. Stiny and Gips (1978) later proposed another category called transparency which refers to cases where the reactions to art works (designs) depend on not only their representations, but also on what the representations may evoke via association with other elements such as emotions, experiences or ideas. Although these discussions were mainly concerned with art work such as music, painting, drawing and literary work, the ideas are also relevant to creative product designs by industrial design engineers. Aesthetics of a product is an intangible aspect which heavily influences the feeling based responses. If we want to link the aesthetics to tangible product characteristics we

need to describe the product in an aesthetic dimension. We have seen that it is possible to formulate aesthetic variables which can span the aesthetic dimension in which the product is represented. However, care has to be taken in working with these variables, because the danger exist that the (subjective) variables represent only a subset of the total aesthetic space and might also be interdependent. In the studies we present in chapter 3 and 4 it is seen, for instance, that two different sets of aesthetic variables are used to describe a product. 2.2 Product characteristics Since product design is a constructive process, we need to examine the characteristics at the most basic level to be of use for integrating aesthetic intents with design. We believe that the three most basic characteristics that influence the aesthetics of a product are: shape, composition and physical attributes. High-level characteristics such as style or fashion may be dealt with by expressing them in terms of these three basic characteristics. Shape, form and geometry are ambiguous terms. Even the Websters dictionary cannot give a precise description of shape and form. Both terms are mixed in the descriptions and used in a tautological way. In the framework of this paper we therefore start with defining geometry as the lowest level explicit description of a three dimensional point set in space. It is in fact the mathematical documentation of points, lines, surfaces, etc. Topology and morphology are placed on the same level as geometry, because they provide us exact descriptions on the structure, and specific properties of geometric elements. We could say that besides spatial specification, geometry has no further contribution and only supports the aesthetic-shape mapping process.
Shape feature Shape Expresses Composition Physical Attributes Aesthetic characteristics Influences Form Triangles

Supports

Geometry

{ (x 1,y1) , (x 2,y2) , (x 3,y3) }

Figure 1

Product characteristics and the interaction with aesthetic characteristics.

One level above the geometry we find form, which is to our definition a categorical representation of global properties of the geometry. It is a more generalized descriptive term. An example of form is a triangle, as it is a category of geometries existing as three interconnected straight lines. The relation between geometry and form is unidirectional in the sense that one form represents multiple (and in fact infinitive) geometries but a

geometry has only one form. In the case of a product form typically influences the aesthetic characteristics. Shape is seen on a higher level than geometry as well. But in contrast to the global characteristics, shape is defined as the totality of local characteristics of the geometry. So shape is an abstract generalization of the local geometric properties. The local geometric properties are shape features which form the basis for shape manipulation on a semantic level. An example of shape is the star-like object in Figure 1. This object is a set of sharp edges, the local shape features, which indeed make the object a star. In relation to aesthetic characteristics the role of shape is to express them. Referring to the above definitions composition expresses how shape features are arranged and therefore act on the same level as shape with respect to the aesthetic characteristics. Guidelines for good compositions which have been recommended by artists and designers for attaining visually pleasing and interesting objects (e.g. Ruskin, 1971), may be deployed for analyzing and comparing aesthetic characteristics of design. Physical attributes such as colour, texture, lighting conditions or material properties also influence aesthetic characteristics, in a similar way as shape does. In fact they add different attributes to the shape and contribute to the total impression of an object. Typically, many aesthetic descriptive terms are shared by more than one of the three main characteristics, with distinct meanings. For example, both shape and colour characteristics may be described as being `harmonious. Some physical attributes also depend directly on shape characteristics (e.g. color appearance), hence cannot be examined in isolation from shape. The main difference to the before mentioned aesthetic characteristics is the tangibility of the product characteristics which provides for a formal description in computer algorithms. As stated before it is the geometry which makes shapes tangible and supports the formalization process. Although the terms to describe aesthetic characteristics are rich and complex, we believe that the essence of these terms can be related to the three basic characteristics of objects, individually or in combination. Since our aim is to provide a general methodology to include aesthetic aspects into product design, it seems to be practical, for the time being, to concentrate on how these basic characteristics of a product influence the appreciation of its aesthetics, and to leave out other aspects. 2.3 A brief review of the past research A survey of literature has revealed that although researchers have analysed the concept of aesthetics for over three hundred years, these efforts generally were concerned with music, painting, drawing and literary work. Very few attempts were focused on design (e.g. Pye, 1995), and even less explored the relation between aesthetics and shape. In psychology and philosophy many studies have been focused on emotions and feelings (Frijda, 1986). In marketing research many studies try to explain consumer behavior, develop models of customer (subjective) judgments (Snelders, 1995, de Bont, 1992) and reliable ways to gather marketing intelligence to incorporate in product design. In civil engineering several attempts have been made to develop systems to design aesthetically pleasing bridges and dams (Furuta, et al., 1993, 1995, Reich, 1993, Miles et al., 1993). In the realm of industrial design Claessen (1996) has explored the relation of color and shape. Kurango, et al. (1992) developed a system to obtain a computer model from

sketches or mock-up 3D-models. Wallace and Jakiela (1993), Takala and Woodward (1998) and recently Hsiao and Chen, (1997), have suggested computer oriented methods that help develop products by using aesthetic intentions or sketches and develop these into detailed designs, using knowledge of customers, design processes and, to some extent, knowledge of translating those into new products. Our literature survey shows that there exist several alternatives of inclusion of aesthetic aspects into the design process of consumer products. A part of the research work tried to include psychological aspects (e.g., observation, feeling) to connect shape and aesthetics. The other part endeavors to explain the phenomena of raising emotions towards a product by assuming a physically determined relationship between aesthetic features and the design variables governing the geometry. From a design methodological point of view, these can be distinguished as phenomenological approaches and systematic approaches. Phenomenological techniques work on a intuitive level and do not strive after a deep understanding of the role of shape features in controlling feelings evoked. On the other hand, systematic techniques tackle the problem penetratingly, rather than superficially only. The objective is to form a methodology based on scientific exploration, procedural formalizing and involvement of computing. A common feature of both intuitive and systematic methods is applying experiments to consumers, designers, products and processes. The reason of the existence of these indirect and direct methods is the lack of understanding of and agreement on what is the intrinsic relationship between aesthetic features and design parameters of products. In short, no objective method of managing aesthetics is known since aesthetics proved to be a subjective, relative and phenomenological category. Nevertheless, researchers have been making efforts for unification of aesthetics and design both in theory and in practice.

3 An example of the phenomenological approach


3.1 Fundamentals of the study Desmet (1998) uses theory from psychology of emotions to model user responses to products, and identifies possibilities for designers to influence those responses with the design of products. The goal is to help design products with an "emotional benefit" and is illustrated in the field of mobile phones. The study is based on a theoretical model of how peoples "feeling based responses" emerge, and clarifies the relation between feelings and products. Based on the theoretical considerations Desmet proposes the following phases in product development to design products with an emotional benefit: Determine customers' profile(s) of initiating and additional needs and their resulting expectations. Do these conflict or correspond with meanings of the product, and what are the functional attributes desired by the customers (-groups). Make categories of corresponding customer profiles. Determine the intended "feeling based response" of every customer group. Determine how these responses are based on product characteristics. Incorporate these product characteristics in requirements for new designs.

The study shows how the so-called expression profiles of two groups of customers were determined and translated to product requirements. Based on this data new designs were proposed and responses to them were measured 3.2 Determination of the target user profile To determine customer profiles Desmet studied (based on Reynolds and Gutman, 1988) how the general meaning a product expresses, acts as stimulus on three levels: attributes, consequences and end-values. In this study mobile phones are analyzed, and customers are interviewed with laddering techniques to determine what structure of meaning these products have for a group of customers. The study resulted in categorization of customers into two "care styles": Trend Followers and Security Seekers. These two groups of people have distinct expectations. Trend Followers want a product to impress others, want to balance their patterns of spending money, time and energy, and want freedom of movement. Security Seekers expect their products to provide calmness, no stress, they do not want to attract attention, want to help people in distress, and want to maintain a good relationship with other people. A common expectation is that they also want mobile phones to comply with their search for a balance in spending money, time and energy. The found expectations together with remarks during the interviews lead to: relevant initiating and additional needs desired consequences of using a mobile phone lists of functional attributes of mobile phones, concerning memory, batteries, design, size, range, user interface and costs. Expectations to be expressed by product characteristics: Calmness, no stress; unobtrusive Profile of expressions sober professional business-like reliable pleasant easy Table 1 Shape characteristics block like; no curved surfaces, no large fillets subdued surface; colors black or dark gray no small parts, robust i.e.: nothing pointing out, no sharp edges large display; few keys, with one clear function, in rows and columns

Care Style: Security Seekers

Example of the expression profile and related shape characteristics used to generate concepts for new mobile phones.

The design the Security Seekers look for in a mobile phone is restraint, sober, pleasant with a business-like, professional expression. The Trend Followers want an eyecatching, cunning design, looking both professional and whimsical, with lots of fun and contrasts. Both groups want small size products each for their own reasons: Security Seekers for reasons of unobtrusiveness and discretion, to be able to carry it anywhere; Trend Followers for making a better impression because small is more expensive looking.

distressed annoyed fearful nervous jittery anxious activated unpleasant affect unhappy miserable sad grouchy gloomy blue

aroused astonished stimulated surprised active intense

enthousiastic elated excited euphoric high activation lively peppy activated pleasant affect happy delighted glad cheerful warmhearted pleased

unpleasant

pleasant

inactivated unpleasant affect dull tired drowsy sluggish bored droopy

Low activation inactivated pleasant affect relaxed content at rest calm serene at ease

Finally, to make the customers expectations operational, two Figure 2 The circumplex of affect, a space with the dimensions "activation" and "pleasantness expression profiles were assembled compiling terms that summarize what the products should express with the help of terms the customers used (Table 1). The shape characteristics are determined in an experiment in which both care style groups were confronted with eight existing mobile phones. The experiment lead to a selection of a feeling based response based on measured feelings and a ranking of the products. The response to these products was measured with an instrument based on the work of among others Izard (1972), who distinguishes the dimensions "Activation" and "Pleasantness" to characterize feelings. The instrument uses non-linguistic means for subjects to express their feelings about a product (see Figure 2, Source: Larsen and Diener, (1992), based on Russell, (1980), and Watson and Tellegen, (1985).). The results of this experiment was that Security Seekers had a pleasing calm feeling with the product of their preference, whereas Trend Followers had a pleasing aroused feeling. These responses were taken as the feelings that new products should evoke in the target groups. 3.3 Translation of intended feelings to and appraisal of the new designs Desmet concluded that, because the concerns and intended feeling based responses are different for the two care style groups, no single product can be developed to produce an emotional benefit for people in both groups. He continued by using the extended expression profiles, together with (designer specific) collages to design some 26 concepts of new products of which he selected 8 to progress with. These were developed into four shape alternatives: (see Figure 3). Two were developed to evoke intended

quiet tranquil still inactive idle passive

feelings: Logica to evoke a pleasing calm feeling in Security Seekers; Helix to evoke a pleasing aroused feeling in Trend Followers. The other two were added for validation purposes, and were designed to evoke pleasing aroused feelings in Security Seekers (Argus) and pleasing calm feeling in Trend Followers (Luna). The translation of the selected intentions to a new product design was done by using the shape sensitivity gathered by performing the research, using collages and Desmets design skills to generate ideas and concepts to Figure 3 select for continuation of the study. This is the most intuitive part of the process.

The newly designed mobile phones

The designed products were then presented to the subjects who again expressed the feelings they experienced using the circumplex of emotions. Additionally the subjects stated their relative preference for the designs, and gave a judgment on the issues in the expression profiles. The analyzed data resulted in an overview of the responses in the circumplex of feelings, an overview of the ranking, profiles resulting from the judgment on aspects, and a factor analysis of the data. All analyses resulted in the conclusion that the framework developed lets the designer influence the feeling based response of respondents by means of the shape of the product. The study has not explained however what is to be done with a different product and a different cluster of users.

4 An example of the systematic approach


4.1 Fundamentals of the study In this section we present a systematic approach that is also intended to link aesthetics aspects and geometry. The work is extracted from the BriteEuRam project on Innovative Styling Applications in Computer Aided Environments (INSTANCE). The basic aim of the project was to investigate and compare expression and (perceived) geometry for both stylistic sketches and the resulting products in automotive design. Muller (1997) already argued that for instance the division of space between the wheel, the driver seat and the position of the engine, in the design of sports cars, evokes certain feelings. He illustrates that a symmetric space division radiates peace and balance while an asymmetric division leads to feelings of dynamics. However, despite the arguments no formalised methodology is mentioned to link these aesthetic characteristics with the definition of geometry. The scope of this study is limited to the aesthetic-to-shape relation. A subset of the aesthetic characteristics is compared with perceived geometric data. In this way we are able to discuss about part of the relation from aesthetic characteristics to shape or the understanding how aesthetics evokes feeling (the upper path in Figure 8).

10

The representative sample in this study was taken from 5 different automotive projects and consisted of a concept sketch and a photograph of the finished prototype (see Figure 4). Six subjects (3 automotive designers, 3 industrial designers) were involved for judging the material. The two subject groups, five different car designs and also the documentation of the initial product idea and prototype of a single car could be compared. This allows for data interpretation along the three dimensions: subjects, designs and product development phase. Among the challenging questions that were to be answered in this study were: Figure 4 Concept sketches and prototypes that were used as stimuli. (a) What is the relation between expression and geometry in automotive design? (b) Can we explain differences in expression by those in geometry? (c) How does expression/geometry propagate through a design process? (d) Is there a difference in the interpretation of two different groups of designers? 4.2 Measuring aesthetic characteristics The aesthetic focus of the study was the perceived expression by the subjects, which was broken down into four terms: aggressive, friendly, functional and elegant. These terms are frequently used in automotive industry and cover most of the expressive qualities of cars, although questions can be raised about the independence of these subjective aesthetic terms and the completeness of the description of the aesthetic dimension by these variables. In reference to the aesthetic dimensions which were mentioned in section 2.1 we think that the terms used in this study represent a subset of the complete space of aesthetic characteristics. Each subject was confronted with the stimuli and appointed a value between 0 and 4 for each expressive term, in which the following meaning was given:

11

0 1 2 3 4 -

not at all aggressive, friendly, functional or elegant not very aggressive, friendly, functional or elegant neutral quite aggressive, friendly, functional or elegant very aggressive, friendly, functional or elegant friendly sketch 0.2 1.8 1.5 2.0 0.8 product 0.8 2.7 2.2 2.7 0.7 functional sketch 0.0 2.0 3.7 2.8 0.5 product 0.7 2.8 3.8 3.2 1.3 elegant sketch 2.0 3.2 3.0 2.0 2.7 product 1.8 3.2 0.7 1.5 1.5

aggressive sketch car1 car2 car3 car4 car5 3.7 2.5 2.2 2.8 3.7 product 3.2 2.3 0.5 2.3 3.3

Table 2. Average results for the expressive terms for all stimuli. For each expressive term table 2 gives the judgment for the sketch and the product prototype of each car. The values are the averages for all subjects. All car sketches are perceived more aggressive and elegant while the prototypes are experienced as more friendly and functional. We will further discuss the values of table 2 in section 4.4. 4.3 Determination of the product shape The shape of the product in this study is practically the perceived shape. For both the stylistic sketches and the photographs of the prototypes the perceived surface is measured by the method described by Koenderink, et al. (1992, 1995). Basically they developed the method to extract three dimensional surface information from the perception of two dimensional images. In this method a subject is sitting behind a computer monitor which displays an image of the stimulus. By adjusting a pushpin like cursor, which is projected on the picture, the subject is able to indicate the local orientation of the surface (see Figure 5). The subject is asked to adjust the pushpin like he thinks the disk of the pushpin lies on the surface. The orientation is described by the slant and tilt of the surface normal. Slant is the degree to which the pushpin leans backwards, and tilt is its wind direction like a compass needle. The mouse serves as a tool to position the pushpin. Figure 5 When the subject is ready positioning the pushpin it is indicated by the mouse as well. The pushpin

An example of different pushpin positions on a surface

12

will now be positioned the next point on the picture. The positioning is done on a (normally invisible) triangular snap grid. After the evaluation of the picture the raw data of slant and tilt for every vertex is computed into a local surface normal. For every pair of vertices a local depth is computed, while the final vertex depth value is estimated by a curve fitting procedure using a least squares criterion where the Figure 6 Triangulated image and 3D reconstructed surface for mean depth value is one of the stimuli. set to zero. Now for each (x,y) value of the grid a depth value z can be added and a 3D-reconstruction of the perceived surface made. In this study all images were triangulated and evaluated with the Koenderink method by the subjects. In this way perceptual data on 60 surfaces was obtained. As an example, Figure 6 shows how the stylistic sketch of car1 was triangulated and how one of the subjects saw its shape. 4.4 Comparison between tangible shape characteristics and intangible emotions After the investigation of the aesthetic feelings and the documentation of the perceived geometry the data was compared across the subjects and the stimuli. As an illustration we will elaborate on the findings on elegance. On average the subjects felt that the prototypes were less elegant than the sketches. For instance for car4 the rating for elegance drops from 2.0 to 1.5. As an illustration both perceived shapes of this car are presented in Figure 7. The geometry differs in the sense that the connection between the hood and the windscreen is experienced sharper in the prototypes. This is indicated by the line drawn along the meshed surfaces. The other cars showed the same characteristics so it was concluded that an edgy surface as experienced in the prototypes gives a less elegant object. Furthermore, it was found that sketches were experienced as more stretched or directed. This was indicated by the occurrence of pronounced long shapes in the surface meshes.

13

The previous example shows that there must exist a relation between expression and perceived geometry (question a). The method used is systematic but more data is needed to generalize conclusions. The judgment of shape characteristics was based on subjective qualities like: the surface is stretched or the surface is edgy. However, it seems to be possible to find some handles to cope with these subjective qualities. An interesting future issue is to look at a comparison of the perceived shape of the image of the product prototype and the actual geometry which of course is known. In this way we can extend our understanding of the aesthetics towards the space of design variables (Figure 8). Figure 7 The perceived shapes of car4 for the This study does not answer the sketch and the product. second question (b) because different designed geometries to measure the intended aesthetics were not used like in the study in chapter 3. A difference in the expressive judgment between the sketches and the prototype (question c) was observed. Geometry tends to be more edgy later on in the process. One of the explanations in this specific field is the enormous amount if engineering restrictions that has to be taken into account during the product development process. Packaging restrictions and manufacturing issues might form a major cause for the change in geometry. This can also be seen in the expressive term whereas the functionality shows to be higher in the final prototype in which the functional requirements are integrated in the design. Finally, some shape interpretation differences between both groups of designers (question d) were seen. Especially, the use of reflections in stylistic drawings, which are used to exaggerate certain shape intentions, lead to more articulated surfaces for the automotive designers. In the context of this paper we have seen that several efforts already have been made to investigate the link between aesthetic and product characteristics. Referring to Figure 8 we can position the research presented in chapter 3 as an approach to investigate the complete loop of understanding how aesthetic evokes feeling and how to design aesthetically pleasing products. This was illustrated using a phenomenological approach. Chapter 4 elaborates on the upper path only in Figure 8 of understanding how aesthetic evokes feeling. Nevertheless, in the described research a systematic scientific method was used giving us more formal information about the phenomena than the extended expressive profiles as used in the study of Desmet. The next chapter will build on these

14

experiences and propose a general methodology for linking aesthetic characteristics and shape.

5 A methodology for unifying aesthetics and design


5.1 The need for knowledge intensive support of aesthetic design Although its importance is well known, formalized and systematized manipulation of aesthetic concepts is almost in its infancy. In commercialized CACD/CAD systems, support of aesthetics stopped at the level of fast shape prototyping and of expressing visual intent by manipulation of shape (surface) features. What we miss is and processing the aesthetics related design knowledge. Present feature technology is closely linked to conventional geometric modeling engines, therefore, involving aesthetics knowledge explicitly is very difficult, if not impossible. The most intrinsic issue of development of a design for aesthetics methodology is how to implement the mapping between the space of aesthetic features and the space of shape-induced geometric parameters. The relationship between aesthetic appreciation and design is rather vague. It also raises philosophical issues about the relation between human beings and artefacts they create. We accepted the fact that any aesthetic judgement is necessarily a potential feature of the way in which we attend to objects (Palmer and Dodson, 1996). Thus the correspondence between the elements (groups) of the two spaces is vague and is not straightforward. It can be apprehended through the involvement of psychological aspects only and it is strongly influenced by cultural, personal, sociological, etc. facts. Aesthetics is a simultaneous communication of meaning and beauty. Therefore, the theory of design for aesthetics has to explain how meaning and beauty are communicated by a particular manifestation of an objects. The principal medium of communication is the shape of an object, but the colour, texture, material and other visual properties play also important role. Actually the totality of these object properties makes impression on us and exerts emotion. This fact makes the task complicated. That is, aesthetics of objects has to be addressed individually. 5.2. The fundamental concept: the analogy of information communication Prior to the development of a computer-supported design for aesthetics methodology we have to model this specific communication process. On Understanding the one hand, because the how aesthetic {F } principles of general evokes feelings 1 {V } 2 information communication {F } 4 {V } are rather well known and, on {F } 1 2 {V } 3 the other hand, since an {F } 3 analogy with communication {V } 4 Designing of aesthetic intent can be aesthetically pleasing products observed, in finding out the Space of aesthetic Space of design characteristics variables knowledge processing fundamentals of the needed Figure 8 The problem of two ways mapping

15

methodology we can rely on the theory of communication in information technology. In communication of information the semantic content is carried by digital/analogue signals. In communication aesthetics the meaning is delivered by shape properties. This information is however implicit and the meaning cannot be completely separated from its interpretation. Consequently, aesthetic designers first have to understand how shape evokes feeling in the case of a particular product and a cluster of consumers. It means that, contrary to the one directional methodology of conventional design for X techniques, design for aesthetic manifests in a closing loop (Figure 8). Our investigation revealed that the loop unifying aesthetics and design, is formed by the linked activities of understanding how a particular shape of a given product evokes feelings, and what designers should do to be able to communicate an aesthetic message and achieve emotional satisfaction and attraction to certain shapes. From the theory of communication it is well-known that four levels can be identified in the information transmission process. These are (a) statistical, (b) syntactical, (c) semantical and (d) pragmatic level. None of the mentioned levels is able to provide effective communication in itself. On the statistical level of communication, the amount, magnitude and distribution of physical signals are dealt with. On the syntactical level the patterns of the signals are extracted and interpreted. On the semantic level the meaning of the patterns of signals is reasoned out. Finally, on the pragmatic level the context and the goal of the communication are interpreted. In addition, recent studies circumscribe a fifth level that is orientated to explain what the motivation and the behavioral background of the sender was. Although not explored exhaustively, the fifth level, that is called apobethical, might open up new dimensions in understanding the deepest knowledge related to communication of aesthetic properties. 5.3. Understanding aesthetics for shape design The analogy of communication facilitates mapping between the space of aesthetic features (evoked feelings) and shape parameters. In the looping process formed by the activities that are targeted to explore and understand why and how does an object evoke feelings and by those that are to design a product achieving a conceived aesthetic impression, similar communication levels. The contents of the levels however differ depending on which side of the loop they refer to. Figure 9 shows the communication{F } 1 {F } 4 {F } 2 {F } 3 Process of understanding how aesthetics evokes feelings Space of aesthetic characteristics Space of design variables {V } 1 {V } 4

{V } 2 {V } 3

Experiment with typical objects

Clustering objects based on feelings

Exploring common characteristics

Identifying shape characteristics

Understanding the influence of shape

Statistical level

Syntactical level

Semantical level

Pragmatic level

Figure 9

The communication model of understanding how aesthetics interacts with people

16

based model of understanding how aesthetics interacts with people. The communication analogy also supports setting up an experimental process to explore and systematize the related knowledge. In order to understand how meaning and beauty interacts with people an experimental sub-process is executed. A representative set of objects of different observable aesthetic characteristics is put together. Potential consumers/users are asked to select those objects that, due to their aesthetically similar or resembling shapes, can be sorted into the same cluster. Only a fuzzy clustering is possible since the objects can be sorted into more than one cluster based on their aesthetic characters of different dominance. Fuzzy clustering is one of the conceived advanced modeling techniques that are useful for data classification and pattern recognition problems. It is an iterative technique that seeks to compute a membership function that indicates to what degree a data point belongs to a given cluster. It can be used either for data analysis or aesthetics pattern recognition. The activities mentioned above are related to the statistical level of communication. The next step is the exploration and naming of aesthetic characteristics common for the objects in a cluster. The users are asked to circumscribe the aesthetic properties they observed and recognized as well as to depict the feelings the objects individually evoked in them. The verbal descriptions of the observations, impressions and the feelings bring us to the syntactic level of communication. The specifications of the corresponding aesthetic properties form the perceptual part of the vocabulary of design for aesthetics. The specifications are based on linguistic variables that express a kind of common understanding of the aesthetic characteristics. The semantical level of communication relates the explored aesthetic characteristics to the shapes of objects and analyses the correspondences. Only the most dominating aesthetic characteristics are taken into consideration and it is investigated which shape features might result in that particular characteristics. This is repeated for all of the clusters in the same way with the same objective. This way clues are formed for the possible relationship. Since the shape is influenced by the less dominant aesthetic characteristics too, formulation of relations cannot be deterministic. On pragmatic level the adequacy of matching the aesthetic characteristics with global and local shape properties is tested. The shape properties are modified and the potential users are asked to judge whether the previously recognised aesthetic characteristics become stronger or weaker, or remain intact. It is also investigated if any of the previously less dominating characteristics change significantly. Although the aesthetic factors were implicitly treated, we can conclude a synthesized knowledge about the influence of the shape for the designers. This instituted knowledge can be formalized in the knowledge base of a computer-aided design for aesthetics system. 5.4. Aesthetics stimulated by shape design Above we described the explorative sub-process that is orientated to understand how shape influences feelings. Figure 10 shows the creative part of the looping process that relates to systematically designing aesthetically pleasing products. In principle it is directed from the space of design variables to the space of aesthetic characteristics. As a starting step, designers should generate an initial shape that fulfils functional requirements. The initial shape is depicted from geometric, topological and morphological point of view. This specification of the initial shape belongs to the

17

Pragmatic level

Semantical level

Syntactical level

Statistical level

Contolling by user experiments

Finishing the product image

Designing local shape features

Designing global shape features

Designing of the functioning shape

Process of designing aesthetically pleasing products {F } 1 {F } 4 {F } 2 {F } 3 Space of aesthetic characteristics Space of design variables {V } 1 {V } 4

{V } 2 {V } 3

Figure 10 The communication model of how aesthetics can be expressed trough design statistical level of communication. Then the shape is globally crystallised to come up with the needed observable shape characteristics (e.g., type, extent, proportions, morphological articulation, distortion, number of components, etc.). Global shaping leads us to the syntactical level of communication since it further specifies the meaning but it does not make it complete. On this level of communication additional design aspects (e.g., material, colour, texture, brandmarks) are also taken into consideration. On the semantical level, the shape designing process manipulates local properties in order to incorporate subject related aspects (e.g.,, style, fashion, human preferences, etc.). As a part of the semantical communication the additional design aspects are also manipulated to achieve emotional impacts. On the pragmatic level users/customers opinions are collected to measure satisfaction and to further improve aesthetic impressiveness. This advance measuring of the evoked feelings however backtrack us to the explorative sub-process which aspire to explain "feeling based responses" to products. Our initial experiments and tests have shown that the information communication analogy helps in understanding the core problem addressed in this paper and supports the elaboration of a quasi-formalized methodology that can be used in the future as the base of tool development to model shapes which a certain aesthetic intent. 5.5 Toward a formalism for the mapping between shape parameters and aesthetic characteristics As mentioned in section 5.1, the correspondence between the space of shape-induced geometric parameters and the space of aesthetic features is vague. However, we argued that an implementation of a mapping between the two spaces is a prerequisite for any methodology of design for aesthetics. The development of such a methodology involves (1) obtaining the mapping and (2) employing the mapping as part of the methodology. Idealistically the mapping specifies, for an expressed aesthetic intent, those values of shape parameters as to achieve a design model conform the intention. We will introduce a set-theoretic description of such mapping in order to investigate the feasibility of the steps (1) and (2) mentioned. A set-theoretical formalism seems very appropriate to

18

clarify such fundamental issues as object identity, context specification and relationship between objects. The two aforementioned spaces will be referred to as the space S of shape variables and the space A of aesthetic variables. We assume that at some point in time, a finite number of shape variables si and aesthetic variables ai are relevant. These variables are contained in the sets {s1,s2, , sm} and {a1, a2, , an} where each of the elements specifies (1) a type and (2) a value domain. We give three examples of possible shape variables: s1 has type height and value domain +, s2 has type angle between arm and body and value domain {zero, sharp, wide} s3 has type y-co-ordinate of the 23d control point and its value domain is .

is a fundamental issue to be discussed later. The value domains may be continuous, discrete, ordinal, nominal or even arbitrary. The types are assumed to capture sufficient semantics. As is common in verbal communication, a label such as height conveys all necessary information in some context. The proposed formalism, however, will mainly be based on the value domains of the variables. Two examples of aesthetic variables are, a1 has type harmony and value domain {-2, -1, 0, 1, 2}. a2 has type emotionality and value domain {sad, angry, joyful, serene}

We define the space S as the Cartesian product of the domains of s1sm. Similarly, A is the Cartesian product of the domains from the shape variables. Let d denote a description of a design object; d is in general a member of some Universe of Discourse D, where we assume that d D is accepted, by some people at some point in time, as representing the design object at hand. We are interested in the relation between a given design model d and points in S and A, i.e. we wish to determine the subsets RS D S and RA D A, where denotes the Cartesian product. These RS and RA specify which properties (in terms of si and aj) apply to a given d and, conversely, which objects in D comply to specified tuples of S and A. It is useful to consider whether or not each of the following statements apply to the relation RS: 1. For given d D there exists no s S such that (d, s) RS (S is incomplete) 2. For given d D there exists exactly one s S such that (d, s) RS (S is unique) 3. For given d D there exist multiple s S such that (d, s) RS (S is ambiguous) 4. For given s S there exists no d D such that (d, s) RS (S is too rich, or D too poor) 5. For given s S there exists exactly one d D such that (d, s) RS (S is deterministic) 6. For given s S there exist multiple d D such that (d, s) RS (S is nondeterministic)

19

The same evaluation can be made for relation RA.

A parametric modeling system typically has RS with property 5 for all s S. However, A S F F property 2 needs not to apply; it is common that a given model is achievable using different values for the variables. If a designer (or a system) fails to produce a model for a A S f particular m-tuple of S then RS has property 4. Let us assume, for a while, that RS has Figure 11 In the most simple property 2 for all d D and additionally has S situation the mapping f : property 5 for all s S, i.e. there R is a S AS is equivalent to (FS)- bijective function F : S D. Similarly, let us 1 assume the existence of a surjective (but FA. possibly not injective) function FA : A D, acting as a candidate relation RA. The mapping f: A S from aesthetic variables to shape variables can then be defined as f(a1,,an) = (FS)-1 FA(a1,,an) (1)

(see Figure 11). For convenience of notation we write a point in A as the n-tuple (a1,, an), where ai is understood to denote a value, rather than a type and a value. Similarly, a point in S is written in the form (s1,, sm). The interpretation of equation (1) is as follows. For any instance of the aesthetic variables we can identify some design model, which, in turn, is specified by a unique instance of the shape variables. This procedure obviously has a very restricted extent. However, it is possible to relax some of the restrictions as we will show later. In general, for given f, each of the shape variables si defined by f(a1,,an) = (s1,,sm), is a function of all aesthetic variables, hence we should write f(a1,,an) = (f1(a1,,an), , fm(a1,,an)) First we examine what the implications would be of the mapping f defined in equation (1). These include: 1. Each point in S-space is associated to exactly one design model in D, 2. Each point in A is associated to exactly one design model in D, 3. Each design model in D has exactly one associated point in S, 4. Some design models in D may have no associated point in A, 5. If two design models in D are different, so will their associated points in S, 6. If two design models in D are different, their associated points in A may be both nonexistent, one may be nonexistent, or both may be existent and be the same or different, 7. Each point in A maps (using f) to one point in S, 8. Two points in A may map to the same point in S,

20

9. There may be points in S that cannot be a result of the map from A. The limitation expressed by items 4 and 9 reflect the scope of equation (1) which does not deal with design models that cannot be prescribed by aesthetic variables. Indeed the primary interest is the capability of f to be map from A to (not necessarily onto) S; this leads to Observation 1: A mapping f from A to S needs not to be one-to-one. This implies that, despite the restrictions we introduced, the specifications using terms in one space are not equivalent to specifications in terms of the other space. Another important observations is as follows: Observation 2: The specification of a mapping f from A to S involves the existence of at least one member dD.

This may be not obvious from equation (1), where the dependence of f on d D is implicitly defined by the relationship RA between A and D. Observation 2 suggests that it is unfeasible to define a dependence relation between, for example, sharpness and agressiveness without reference to at least one object. It is important to verify whether or not these observations depend on the restrictions that we introduced for equation (1). Before doing so we consider the special case, where the shape variables and the aesthetic variables form real vector spaces, i.e. S = m and A = n. For this case we define the Jacobian matrix Jf of the map f : n m as the mn matrix of the first derivatives of f, i.e.: Jf(a1,,an)ij = /aj fi(a1,,an). The Jacobian matrix is commonly used to analyze maps between two vector spaces that are both domains to represent kinematic systems (Latombe, J.-C., 1991). The entry (i, j) of Jf specifies how much variable sj changes due to changes of variable aj. In accordance to observation 2, this change is dependent on the point in A where it is evaluated. In kinematic analysis the function f is sometimes reconstructed from measurements of its Jacobian matrix, which is a principle that may carry over to our field of interest. 5.6 Obtaining a map f and applying it to design for aesthetics To specify a map f we need to know the structure of the spaces S and A, i.e. the types and the value domains of their variables must be defined. Let us review what has been achieved, in this respect, in the study reported in chapter 3. In section 3.2 the types intensity and pleasantness, among others, were introduced and their value domains specified in figure 2. Subjects were confronted with eight existing mobile phones in order to determine design model-dependent values of the variables. This can be regarded as an attempt to establish the relation RA between the spaces D and A (if we consider intensity and pleasantness to be aesthetic variables). This relation was found to be subject-dependent. In section 3.3 variables of S were correlated to elements from D, which can be regarded as a search for the relation RS. Next, a positive correlation between some of the si and aj was reported based on statistics induced by stimuli, i.e. elements of D. Finally, this correlation was applied to create new elements of D (i.e. new designs) as follows. If it was found that for relatively many elements in D a value i of variable ai correlated with value j of variable sj, then it was hypothesized that a

21

new q D for which si = j will also have ai = i. This expectation was then verified, as described in chapter 3. The aforementioned positive correlation can be regarded as an indication of the existence of a map f, although the data allow, at most, the specification of a relation between S and A, rather than a mapping. This interpretation of the report in chapter 3, as well as more general considerations raise several issues: 1. The choice of types of variables ai and si seems to be inspired by the design objects di at hand. This limits the range of any relationship RA and RS to a particular subset of D, which may be too restrictive if the relations are to support design for aesthetics. 2. It is important to distinguish between the type and the value of variables; in table1 the shape characteristics seem a mixture of types and values. 3. The relationships RA and RS should be based on empirical data. Subject dependence can be incorporated by either introducing subject-specific (or subject categoryspecific) relations, or by including ome subject-dependent weighting function. 4. Measuring RA and RS is not a well-defined process. The assignment of values to the shape and aesthetic variables is mostly based on panel responses and is unlikely to be a reproducible action. This instability has repercussions on the mapping f. To gain stability (c.f. issues 2 and 3) one could refer to subsets of D rather than individual elements when establishing the relationships RA and RS. Indeed, when a specific design di D empirically relates to some si S then this may be true for some neighborhood of di.. This would require D to have a topology. Conversely, a topology on D could be based on a classification of D, where a class is formed by elements relating to the same point in S (or A). Another approach is to not attempt to directly obtain RA and RS, but to measure their changes with respect to changes in the space D. For example, based on panel responses it seems easier to detect that a certain edge gets sharper, than to obtain agreement that a particular edge is moderately sharp. This is in analogy to measuring the Jacobian matrix of a robot in attempt to reconstruct the mapping from configuration space to Eucledian space. Obviously, the notion of differentiation of shape (or aesthetic) variables with respect to d requires continuous or quasi-continuous properties of the involved spaces. Design for aesthetics could then be supported if the aesthetic intent is specified as a modification (a1,,an) relative some reference design di. Then, if the restrictions for equation (1) apply, the candidate new design di+1 is obtained by changing the shape variables as follows (s1,, sm) = + (a1 Jf(a1,, an)1,1,., an Jf(a1,, an)1,n) + (a1 Jf(a1,, an)m,1,., an Jf(a1,, an)m,n),

where (a1,,am) A represent the aesthetic variables values for di, i.e. FA(a1,,an)=di. The new di+1 is di+1=FS(s1 + s1,,sm + sm). To gain detailed insight in the nature of the mapping f (either the one according to equation (1) or a less restricted relation), we recommend to perform initial experiments

22

using relatively simple spaces S, A and D, primary to test the stability of value assignment to the variables.

6. Discussion and conclusions


There exist uncertainties about the scientific understanding of how aesthetics and shape design meet. The terms used to describe aesthetics are rather subjective and have no direct relationship to the characteristics of the shape as a whole, or as a composition of its parts. Aesthetics is conveyed by the shape which is however only one aspects of materialization of the product. When aesthetic effect is considered as a totality, other aspects like colors, texture, patterns, etc. are to be taken into account too. In this paper the authors attack three issues of design for aesthetics: (a) to understand the way how products evoke feeling, (b) to link product characteristics to aesthetic factors, and (c) to develop a methodology that enables us to provide better computer support to design for design for aesthetics. Several authors claimed that a product gives rise to sensory stimulus, i.e. perception, that in turn brings about feeling-based responses. Therefore, dealing with aesthetics has a lot to do with the psychology of inspired emotions. First of all, aesthetics of a product attracts the user in the shops, advertisements and magazines. But the aesthetics also influences the relation of the user to the product when it is in use. It means that explanation of appreciation need simultaneous consideration of the human behavior as a consumer and a user. While a designed product can trigger definite aesthetic responses to observers, it is not easy to relate these responses to the characteristics of the product. In most of the time, designers work instinctively. They spontaneously apply their skills and knowledge and judge their successfulness in incorporating aesthetic intends based on their own emotions. Efforts have been made to explore and explain the relation between aesthetic intent and design properties, forming a loop between users and designers. However, the methods they use are strongly dependent on the product, people, objectives and situation and are not fully scientific. On the contrary to these phenomenological approaches, some work tried to apply formal approaches and scientifically sound methods. Due to the lack of some of the knowledge needed, but also to the complexity issues, they have been compelled to treat the problem either from the aspect of user-product, or the aspect of designer-product relationship. By giving up the desire to be comprehensive and by focusing particular aspects only, these systematic approaches have not been able to deliver the requested methodological framework for design for aesthetics. The paper gives insights to two studies that supports these statements. In order to create a closed loop between designers and users, or users and designers, and to unify aesthetics and designing, the paper proposes a framework which utilize the analogy of information communication. The communication loop of aesthetic properties divides into two parts: one is an explorative sub-process that is orientated to understand how shape invokes feeling, the other is a creative sub-process that relates to systematically designing aesthetically pleasing products. Although the rich diversity of aesthetic characteristics makes it impossible to produce categories that cater for all tastes and styles, it is nevertheless feasible to identify classes that cover the essence of more commonly accepted aesthetic intents and leave out individualistic differences that

23

are more extreme. The problem of delivering aesthetic intent is decomposed onto statistical, syntactical, semantical and pragmatic levels of communication. Our initial experiments and tests have shown that the information communication analogy helps understanding the core problem of design for aesthetic and supports the elaboration of a quasi-formalized methodology that serves well as a base of computer tool development. The implementation of a more formalized design for aesthetics methodology arises two technical issues: (1) obtaining the mapping procedure, and (2) employing the mapping as part of the methodology. We are developing a set-theoretic approach for the computer supported mapping of aesthetic characteristics to shape characteristics. To gain detailed insight in the nature of the mapping f we are going to perform further experiments and to test the stability of value assignment to the variables.

Acknowledgment
The authors wish to thank Pieter Desmet for his support to elaborate this paper. Cees Overbeeke was so kind to provide the information on the INSTANCE project (BriteEuRam no.3, project 95/2151). The research work reported in this paper relates to the ICA research project of the Sub-Faculty of Industrial design Engineering of the Delft University of Technology.

References
Beardsley, M.C., 1958, Aesthetics, Hartcourt, Brace & World, New York. Bont, C.J.P.M. de, 1992, Consumer Evaluations of Early Product-concepts, Ph.D. Thesis, Delft University of Technology, Delft University Press, Delft. Chen, K., Owen, C.L, 1997, Form Language and Style Description, Design Studies, vol.18, nr.3, pp.249-274. Claessen, J.P., 1996, Shaped by Colour, a Study on the Effect of Colour on the Perceived Shape of Objects, Ph.D. Thesis, Delft University of Technology, Delft University Press, Delft. Desmet, P.M.A., Tax, S.J.E.T., 1998, Designing a Mobile Telecommunication Device with an Emotional Surplus Value: a Case Study, Design Studies (in preparation). Dickie, G., 1974, Art and the Aesthetic: An Institutional Analysis, Cornell University Press, Ithaca and London. Frijda, N.H., 1986, The Emotions, Studies in Emotion and Social Interaction, Cambridge University Press, Cambridge, UK. Furuta, H., Hase, H., Watanabe, E., Tonegawa, T., Morimoto, H., 1993, Applications of Genetic Algorithm to Aesthetic Design of Dam Structures, Advances in Engineering Software, vol.25, no.2-3, p. 185-195. Furuta, H., Maeda, K., Watanabe, E., 1995, Application of Genetic Algorithm to Aesthetic Design of Bridge Structures, Microcomputers in Civil Engineering, vol.10, no.6, p. 415-421. Goldman, 1995, Aesthetic Value, Westview Press, Colorado.

24

Hsiao, S.W., Chen, C.H., 1997, A Semantic and Shape Grammar based Approach for Product Design, Design Studies, vol.18, nr.3. Izard, C.E., 1972, Patterns of Emotions, Academic Press, New York. Kant, I., 1790, Critique of Judgement (Kritik der Urteilskraft). Koenderink, J.J., van Doorn, A.J., Kappers, A.M.L., 1992, Surface Perception in Pictures, Perception and Psychophysics, vol. 52, no. 5, pp.487-496. Koenderink, J.J., van Doorn, A.J., Kappers, A.M.L., 1995, Depth Relief, Perception, vol.24, no.1, pp.115-126. Kurango, T., 1992, FRESDAM System for Design of Aesthetically Pleasing Free-form Objects and Generation of Collision-free Tool Paths, Computer-Aided Design, vol.24. no.11, pp.573-581. Latombe, J.C., 1991, Robot Motion Planning, Kluwer Academic Publishers, Boston. Larsen, F.J., Diener, E., 1992, Promises and Problems with the Circumplex Model of Emotion, Review of personality and social psychology, vol. 13, pp. 25-59. Mijksenaar, P., 1997, Visual Function: An Introduction to Information Design, OIO Publishers, Rotterdam. Miles, J.C., Moore, C.J., Evans, S.N., 1993, Deriving Rules for Bridge Aesthetics, Knowledge Based Systems for Civil and Structural Engineering, p.vi and p.325. Muller, W., 1997, Vormgeven, ordening en betekenisgeving (in Dutch), Lemma, Utrecht. Palmer, J., Dodson, M., 1996, "Design and Aesthetics: A Reader", Routledge Publisher, London. Pye, D., 1995, The Nature and Aesthetics of Design, The Herbert Press, UK. Reich, Y., 1993, A Model of Aesthetic Judgment in Design, Artificial Intelligence in Engineering, vol.8, no.2, pp.141- 153. Ruskin, J., 1971, The Elements of Drawing, Rover Publications, New York (first published 1857). Russell, J.A., 1980, A Circumplex Model of Affect, Journal of Personality and Social Psychology, vol. 39, no. 6, pp. 1161-1178. Snelders, H.M.J., 1995, Subjectivity in the Consumers Judgement of Products, Ph.D. Thesis, Delft University of Technology, Delft University Press, Delft. Stiny, Gips, 1978, Algorithmic Aesthetics: Computer Models for Criticism and Design in the Arts, University of California Press, Berkeley, LA. Takala, T., Woodward, C.D., 1988, Industrial Design based on Geometric Intentions, Theoretical Foundations of Computer Graphics and CAD, Ed. B.A.Earnshaw, NATO ASI Series, vol.F40, pp.953-963. Tolstoy, L., 1960, What is Art?, Bobbs-Merrill, Indianapolis (first published 1896).

25

Wallace, D.R., Jakiela, M.J., 1993, Automated Product Concept Design: Unifying Aesthetic and Engineering, IEEE Comput. Graph. Appl., vol.13, pp.66-75. Watson, D., Tellegen, A., 1985, Toward a Mood,Psychological Bulletin, no. 98, pp.219-235. Consensual Structure of

Wolff, M. de, 1992, Over Henry van de Velde (in Dutch), Volkskrant, Issue on September 11.

26

You might also like