Pirita Seitamaa-Hakkarainen
The model developed in the author’s previous study (Seitamaa-Hakkarainen, 1997a) conceptualizes weaving design on the basis of two problem spaces: 1) a more domain-independent design process in a composition space and 2) a domain-specific design process in a construction space. The selection of the visual elements requires a search through the composition space, and the selection of technical elements requires search through the construction space. The model at a generic level describes how information from each space guides the search in the other i.e., what kind of transfer and transformation happens between the spaces. Analysis of design processes from the viewpoint of the two problem spaces gives new insight and more developed conceptual tools for analyzing the interaction between domain-general and domain-specific knowledge in design processes in general, and weaving design in particular. This section will introduce the theoretical background of the development of the dual-space search model for weaving design. In the section following, the preliminary model will set out in more detail.
Analysis of problem-solving spaces has been very important in recent attempts to analyze complex problem-solving processes (Goel & Pirolli, 1992; Goel, 1995; Goldschmidt, 1997). The original idea of two spaces of human problem solving was introduced by Simon and Lea (1974) as a space of instances and space of rules. Kulkarni and Simon (1990) extended the idea of dual-space search to the domain of scientific reasoning, where the two spaces in question were the space of hypotheses (generation of ideas) and the space of experiments (testing of ideas). They also provided a simulation model of interaction of these spaces in the context of scientific reasoning. In their simulation model (called KEKADA), on the basis of an existing hypothesis (in a hypothesis space) the system chooses an experiment to test the hypothesis. The system also interprets the results of the experiment (in an experiment space) and thence modifies its hypotheses (Kulkarni & Simon, 1990).
Using the same idea, Klahr and Dunbar (1988; Dunbar & Klahr, 1988 involves a corresponding conceptualization in the area of literate expertise; see Bereiter & Scardamalia, 1987; Geisler, 1994) demonstrated, how their subjects were searching in both hypothesis space and experiment space while solving a problem, and explained how the search in one space affected the search in the other. Dunbar and Klahr (1988) asked the subjects to formulate hypotheses about how a new function (repeat key) of a computer-controlled robot tank (called BigTrak) worked and then to perform the actual experiment (program), in order to find out how the new device key exactly represented the repetition command. The use of a thinking-aloud method allowed the researchers to analyze the interaction between the hypothesis-formation and experimental-design phases of the problem-solving process.
Bereiter and Scardamalia (1987) have carried out numerous carefully
conducted studies concerning the process of written composition, by
analyzing the composition processes of children, adults and experts
writers. They have conceptualized two ways in which subjects produce
written composition: knowledge-telling strategy and
knowledge-transforming strategy (Bereiter & Scardamalia, 1987).
According to them, knowledge-telling strategy indicates that one just
tells what one knows about the topic, which is primarily a linear,
nonreflective process. Knowledge-transforming strategy, in contrast,
means that one will simultaneously work with two problem-spaces,
developing content of the composition (e.g., opinions, inferences) on
one hand, and knowledge that is specially tied to text production, on
the other. Accordingly, the dual-problem space model represents
reflective processes in written composition, where content space and
rhetorical space are connected with linking operations (Bereiter &
Scardamalia, 1987). Moreover, the dual-problem space model provides a
plausible way to account for a variety of expert-novice differences in
writing. Novices can transfer information from the content space to
rhetorical space, but lack the return loop; experts, however, carry on
this two-way process, resulting in more coherent and connected
development of the written composition (Bereiter & Scardamalia,
1987). Geisler (1994) has relied on Bereiter’s and Scardamalia’s (1987)
dual-problem space model, in describing the expert’s writing process.
She stated that writing activity must shift between problem space of
domain content and problem space of rhetorical process.
The idea of conceptualizing the process of weaving design as a dual-space search or process was presented independently by Seitamaa-Hakkarainen (1989), although the idea was not completely articulated. Based on her previous findings the present author expanded the analysis more directly to the dual-space model, in order to analyze the differences between experts’ and novices’ weaving-design process. Primarily, the model developed in the previous study (Seitamaa-Hakkarainen, 1997a) characterized design process as either serial processing or parallel processing within and between the two design problem-spaces. Serial processing means that designing happens by considering only one problem space at a time, i.e., a designer first tries to solve a visual problem, and after that moves to the problem of the construction of visual ideas. Parallel processing means that a designer simultaneously considers, in an integrated manner, both visual and technical aspects of a design problem in a given period of time. The interaction between the spaces, moreover, is seen as an iterative and cyclic process. This is partially based on the fact that design elements are specified incrementally during the designing (Abelson & Soloway 1988; Goel & Pirolli 1992; Goel, 1995).
Expertise in weaving design was studied by using two challenging experimental tasks which corresponded to important aspects of meaningful, real-world tasks of weaving design (Seitamaa-Hakkarainen, 1989; 1997a). Analysis of the design process was based on an open-ended design assignment. There were four subjects; two of them were students of textile; the other two were experienced, weaving designers. The verbal protocols were analyzed by using qualitative content analysis, and by producing behavior graphs of each subject’s problem. The process of weaving design was conceptualized as a dual-space search through composition and construction spaces, with respect to specified weaving-design elements.
Analysis of the novices’ design process by using problem-behavior graphs showed novices started to design from the composition space and only occasionally manipulated visual and technical design elements in a parallel way. A very important feature of the novices’ design process was the dominant role of the development of the composition of the textile (i.e., sketchy model). Due to the complexity of the weaving design problems and the many levels of detail that had to be considered, the novices’ designing occurred mostly in the composition space; design of a composition was the novices’ focus until the end of the design process. They scarcely considered construction problems until they had developed a detailed visual frame for the textile. Thus, the novices’ design process focused strongly on development of one particular design element at a time. In the novices’ design process, only the frames of visual designing interacted effectively; there were no effective interrelationships between frames of the technical designing, or between the two design spaces. Toward the end of the design process, the novices moved into the construction space to update their sketchy model into a more operational form.
The iterative nature of the novices’ design process concerned only visual design. The novices did not further consider the visual elements while solving the technical problems. They were not able to question earlier visual solutions in light of the possibilities of technical construction, or otherwise simultaneously consider composition and construction; the technical elements were solved in isolation and serially. In the case of visual designing, even if the design process was internal to an episode, there was considerable carryover and a parallel development of visual design elements (color and pattern and visa versa) reflecting the incremental development of the design. Without cyclical (repetitive) revising and iterative processing, the solution of technical elements in the novices’ design process remained at a rather superficial and general level.
The qualitative content analysis, as well as problem-behavior graphs, captured the serial nature of the novices’ design process. Due to the limits of the novices’ domain-specific knowledge and lack of an iteratively developed understanding of the principles of weaves, they could not control the technical elements nor manipulate them together with visual elements. The results suggest that the novices’ weaving design is predominantly based on the use of the domain-general knowledge of visual composition, rather than domain-specific knowledge of the constitution of weaves.
The experts’ weaving-design process tended to start from the construction space leading to a parallel consideration of design elements within and between the problem spaces. Unlike the novices, the experts did not start by considering one design element at time; rather, the experts connected many of the design elements together from the very beginning. In the middle of the process they did move back and forth, in serial fashion, between design space by designing elements. Yet generally experts’ design processing was iterative and cyclical in nature in both of the design spaces. By starting from the construction space, the experts first developed the operational model and transferred this model to the sketchy model. By revising design elements cyclically and iteratively, the experts incrementally developed the design and effectively carried previous ideas over to another design episode or, more significantly, into the other design space. Thus, the experts were able to simultaneously to carry out incremental development of the operational model as well as the sketchy model. The experts’ design represented a higher level of iterative process.
The working hypothesis in the previous study (Seitamaa-Hakkarainen, 1997a) was that serial as compared to parallel processing with the two problem spaces, i.e., composition and construction spaces, is the most fundamental difference between the novices and the experts respective design processes. The results of the previous study gave strong support for the hypothesis. There were, however, essential differences in the decomposition of design problems between subjects. Indeed, there are many different ways of starting design; the decomposition is partially subjective and reflects the experience of the designer. The relationship between visual and technical elements may vary from one weaving design task to another, and in some cases visual considerations are primary. Yet in the concrete process of weaving the product, the composition and the construction design spaces have to be considered in parallel in order to yield the intended woven product. Moreover, designing, in its initially technical form constitutes the basis of effective interaction between design spaces. Only extensive experience and practice in weaving design provides the expertise needed for design based on parallel, interactive search processes. Therefore, it may be argued that this principal, identified difference between the experts’ and novices designing is likely represented across many if not all weaving-design tasks; as such it is a basis for valuable hypotheses in further research. In the next section, the preliminary model for weaving design involving the dual-space search will be explored in detail.
As stated, weaving design is seen an interplay between composition space and construction space. Further, the model also characterizes the design process as serial processing or as parallel processing within and between those two design problem spaces. The dual-space search model emphasizes a link between the design process and the content of the design; it conveys the relationship between different design elements.
In order to create a model of weaving design as dual-space search, it is necessary to describe how the elements of the composition space are transferred to the construction space. In the following, the basic components of the transforming operations are defined. The visual design process yields a visualization of the weaving product as a sketchy model of the design. Search in the construction space provides a realization of the sketchy model as an operation model of the product. Moreover, the interaction between the spaces is seen as an iterative and cyclic process in nature. This is based partially on the fact that design elements are incrementally specified during the designing (Goel & Pirolli, 1992). Before displaying the model of the weaving-design process, the concepts used in the model will be defined. All concepts and definitions are based upon the theoretical background presented in earlier sections.
Design space refers to the whole design-task environment, which is common to all prototypical design problems (Newell & Simon 1972; see also Goel & Pirolli 1992; Goel, 1995). Generally, the design space is ill-defined in the sense that there are no definite criteria for testing whether a proposed solution is successful or not. Moreover, the design space cannot be defined unambiguously. The designer has to structure and limit the huge problem-space by using external and internal constraints.
Composition space consists of the organization of the visual
elements and principles selected and manipulated during design process.
The visual elements consist of shape design, color design, and pattern
design elements. Construction space consists of organization and
manipulation of the technical elements and principles. The technical
elements include material design, structure design (e.g., weave and
density) and design of production procedures (e.g., technique, yarn
floats). Technical design strongly influences texture surface. A
dual-space search model for weaving design is presented in the following
figure 1:
Figure 1: Dual-space search model for weaving design.
The visual design process yields a visualization of the weaving product as a sketchy model of the design. Search in the construction space yield a realization of the sketchy model as an operation model of the product. Sketchy model refers to the arrangement of visual elements, which provides a hypothetical frame of the weaving product in a visual form. The sketchy model as such does not provide information on how the drawing can be executed by weaving. Operation model refers to the arrangement of technical elements, which provides an increasingly more concrete model of the weaving product. The assumption is that during the design process the sketchy model will be given an operational form, i.e., drawing sketches makes it possible to explicate relations between weaving elements and thereby to imagine how to realize the drawing in weaving. Design episode is a describable segment of behavior associated with attaining a goal. Each episode contains a unique goal (design element) that will be treated and achieved during an episode. When the goal is accomplished or changed the design episode ends. Goal indicates the intentions and the course of action of the designer in dealing with a design element; goal refers to the content of attention
In general, a designer’s knowledge base includes many types of knowledge of both design processes and the products being designed. The knowledge base consists of prior knowledge and design strategies, declarative knowledge of visual and technical elements and principles and, in addition, procedural knowledge about how to apply that knowledge. The knowledge base involves knowledge stored in the long-term memory. Transformation consists of the operators and underlying mechanisms that move the design process. The most important operators to be considered in this study are analyzing, proposing and evaluating.
Weaving design often starts with a vague idea of the desired product. At the beginning of the design process, the design space is not determined in any specific way. By definition, design space, as a whole, contains many kinds of plausible considerations of the external constraints, ideas that designer might at some point consider (shapes, patterns), available materials (wool, linen), a great number of plausible solutions to previous design situations (i.e., experience), and many ideas how the design process may be organized (start with material design). During the process of design, elements in both spaces are manipulated and transformed to a more specified state to create a workable design. During each design episode, a particular design goal is activated, either in composition space or in construction space. The design element is manipulated in this space at an appropriate level of detail. Further, a design element may be connected to other elements in this particular space or developed many times (i.e., search cycles inside each space). Thus design elements may be in interaction; for example, pattern and color have interconnections during a design episode. However, there can also be interactions between elements of the different spaces. For example, a material element may be connected with a weave element during that design episode. In a design episode, the element becomes increasingly better defined in the operation model. It is assumed that usually all elements are considered or manipulated during the weaving-design process, but not in a particular order.