6.2. Supporting Situated, Perspectival, Linguistic Interpretation
The analysis of interpretation in Part I suggests that computer support for design should:
(a) Capture computer representations of tacit situated understanding at the points when it becomes articulated as explicit interpretations.
(b) Provide multiple perspectives for analyzing and understanding designs.
(c) Allow users to evolve and refine interpretive expressions in language without starting from scratch or accepting predefined frameworks.
This dissertation will pursue these possibilities. Accordingly, three hermeneutic principles will be adopted in trying to develop computer-based environments to support the work of designers:
(a) Provide facilities so designers can create representations of the design situation during the process of solving the task.
(b) Provide facilities so designers can define multiple interpretive perspectives on design problems.
(c) Provide facilities so designers can articulate explicit conceptualizations in language expressions for their work and submerge this new knowledge into tacit forms of knowledge for future use.
These principles will be used to select relevant systems from the literature to review. They will also provide a framework for critically evaluating the systems in Chapter 7. Then, in Chapters 8, 9, and 10, the design of Hermes will be discussed. Hermes is a prototype software substrate that extends the functionality of the domain-oriented design environments and knowledge representation languages that are reviewed so they can support interpretation. The three hermeneutic principles will be used to justify the primary features of Hermes :
(a) An extensible computational medium for representing and evolving artifact constructions, design rationale, computational critics, and other forms of design knowledge.
(b) A mechanism for sharing group and personal interpretive perspectives to support collaboration and deliberation.
(c) A language for explicitly defining computations and for hiding information that can then function in a tacit way.
Capturing explicit understandings of the situation. Human cognition is a complicated business. A recent analysis of its structure by Donald (1991) based on anthropological, neurological, and linguistic evidence suggests four stages in its historical development, all of which remain still active in contemporary cognition: (i) episodic memory that is case-based; (ii) mimetic memory that is tacit or gestural; (iii) mythic memory that is social and linguistically founded; and (iv) extended memory of modern thought that relies heavily on using external media such as pictures, writing, and computers. Heidegger's (1927) philosophical analysis of the logical structure of human being-in-the-world can be seen as a parallel to this sequence: (i) there is the preunderstanding of the world as disclosed to us and of meaningful things discovered in the situation; (ii) through interpretation we use our tacit skills to make things stand out as what they are; (iii) discourse allows us to talk about and thereby share things; and (iv) with assertion we can form predications and externalize our knowledge. In both Donald’s theory of the evolution of the consciousness of the species and Heidegger’s theory of the development of an individual’s cognitive acts, increasingly explicit understanding emerges out of and on the basis of primordial tacit understandings that remain active under the surface.
A system for computer support of human cognitive activities such as innovative design should at least take this complexity into account, recognizing the role of tacit preunderstanding at the origin of all true understanding. That is the reason for the person-centered emphasis: to keep the intentionality of people in the decision-making loop. Ideally, one would hope to provide support for the various levels of explicitness that continue to play a role in even the most sophisticated understanding. For instance, (i) one might try to provide a computer representation of case-based memory to stimulate human episodic memory and intentionality. (ii) Direct manipulation of graphical icons might be used as an analog to mimetic understanding. (iii) A language facility for people using the computer system to make assertions about objects in the domain in which they are working could provide the ingredients for mythic-linguistic comprehension of that domain. (iv) Finally, the computer can provide a computationally active medium that can extend human mental abilities by storing and manipulating vast quantities of symbolic information.
The successive transformations of information from tacit preunderstanding to increasingly explicit forms of expression result in codified knowledge, as shown at the bottom of Figure 5-5, above. As Heidegger pointed out, this form of knowledge provides the possibility of formal science. By the same token, it puts knowledge into a state that can be captured for the physical symbol systems of computer representation.
In particular, if a computer system is being used to provide a medium of external memory for people, then the representations they are using in that medium are available for retention by the computer system for future use. This suggests that computer systems to fill this role need to be structured to provide an enticing and useful medium in which people will represent and solve their tasks. They should also be structured to capture representations as they are developed and used for future reuse and modification. Given the cyclic nature of interpretation, this works well, because the logical source of building blocks for representing the situation of a given task is traditional representations used in the past for representing similar tasks. The abstract chicken and egg problem is solved by starting with a seeded (to shift the metaphor a bit) database. The seed is, of course, generated on the basis of (real or imagined) previous tasks in the domain.
In general, what one wants to capture in the database that grows out of the seed is a palette of symbols for supporting the various forms of interpretation used by people working in the given domain. In particular, it is necessary to support the representation of situated understanding. The situation is a network of significance. Computer representations of the situation might include icons with defined behaviors that are semantically meaningful to people who will be manipulating them. It could also include domain terminology—perhaps structured as a semantic network to reflect interrelationships among terms—to help people formulate meaningful assertions about their work. For a field like design, it would likely include tools for sketching and catalogs of sample designs.
If design is a constructing of reality as argued in Chapter 5, then a computer system to support interpretation in non-routine design is a medium to facilitate such construction. The social construction of reality generally—of what we call the “real world” we all live in—takes place largely through the mediation of artifacts. Design as the design of artifacts (e.g., habitats) is therefore the construction of the artifacts through which reality is, in turn, constructed. The computer systems—as media for design—construct a reality in which this design can take place. In designing approaches (e.g., people-centered) to such systems, one is formulating a theory for building systems (e.g., Hermes), for providing representations (e.g., graphical design components, rationale issues), for designing artifacts (e.g., lunar habitats), for mediating the social construction of reality. This is an example of a high-level design task and illustrates the open-ended character of the tasks that human cognition sets for itself and that exceed the capability of the unaided individual mind.
Organizing perspectives on shared knowledge. Complex design tasks require collaboration. The design of lunar habitats, for instance, began in the 1960’s and may continue for one or more additional decades before the first lunar habitat is ever lived in. This design process relies upon many teams of professionals, from numerous fields of expertise. The teams that continue working over long periods of time change their membership. Designs pass from team to team, developing from requirements documents, to design concepts, to drawings for technical reviews, to mock ups, to construction, and so on. The designs change and earlier stages are iterated. Ideas, explorations, and rationale from one team or one design are reused in others or eventually forgotten.
At one level of analysis, the design of lunar habitats is an unfathomably intricate social process involving hundreds of people, shifting political priorities, changing technologies, financial constraints, and management headaches. But at another level, each actual act of designing is ultimately carried out by an individual person. Some concrete individual must make the suggestion that the wardroom table be shaped a certain way. Other individuals must either agree with or argue against that decision. Each time an individual comes up with an idea or considers a proposal, an act of interpretation takes place. Such acts of interpretation are grounded in the interpreting person’s situation, perspective, and language. People are necessarily the cognitive atoms of collaborative design because only individual people can ground interpretation in intentionality. So the overall social process is dependent upon many individual perspectives interacting in a process of deliberation.
A computer system to support such social processes of interpretation, involving deliberation among multiple perspectives, should provide mechanisms that reflect and aid this process. First, designers using the system need to be able to represent their own ideas and their design rationale. If teams of designers are to use the system, then there should be ways for team members to represent their personal ideas, sketches, arguments, and conceptualizations. The representations of one person need to be kept distinguished from those of other designers. At the same time, design is a collaborative process. The purpose of a design team is to share each other’s ideas, to bring different perspectives on a problem together and to arrive at a consensus through argumentation. A system to support this must allow separate definitions, but also facilitate bringing these differing views together and resolving the differences. Consensus or resolution may not always be possible, so one might also want mechanisms for maintaining minority or competing opinions, by means of which it can easily be determined who supports which ideas and why.
The concepts and design rationale for complex design projects evolve. The contents of individual perspectives must be easy to change. In addition the structure of perspectives must be able to evolve fluidly, defining, for instance, what shared group perspectives include which personal perspectives, or include which particular contents of various personal perspectives. It might be useful to be able to chart the history of evolving ideas, establishing “snapshot” versions of designs in particular personal or group perspectives.
The suggestion is that all representations of design knowledge should be stored in system-defined perspectives because all interpretation is situated in perspectives. This notion transforms the idea of traditional knowledge-based systems that there exists a single body of domain knowledge in the minds of domain experts and that there should be a corresponding fixed knowledge representation. Now the knowledge represented in the system is (1) always relative to the selection of a perspective and (2) continually evolving. The design environment must provide tools to support this view of knowledge and to facilitate the evolution of networks of perspectives and of their knowledge.
Linguistic tools for collaboration. Language plays an undeniable role in collaborative design. Even the individual designer engages in discourse when conducting innovative design. As previously discussed, all acts of interpretation essentially involve explication, which depends upon discourse. Design in teams involves the use of language for the deliberation of design rationale from various perspectives. Much of the work of teams takes place linguistically through group meetings and written reports. These media rely on assertion for the communication of ideas. Language forms the basis for the sharing of knowledge, which is the hallmark of collaboration. The formulation of understandings in language makes possible the formalization of knowledge in methodologically-based systems and the representation or capture of knowledge in external media such as documentation.
As seen in the videotaped lunar habitat design sessions, the design process relies heavily upon concepts, rules of thumb, and constraints. Each of these can be more or less formulated in sentences. To some extent they can even be codified, formalized, operationalized, computerized; to some extent they remain tacit, out of either necessity or practicality. In providing computer support, it is important to analyze the use of techniques like conceptualization, rules of thumb, and constraint formulation. One should support the explication of these to enhance their sharing, retention, and reuse. At the same time, one should recognize that the advantages of explicit knowledge are offset by costs in time, cognitive effort, rigidification, and loss of intuitive control. Designers may wisely decline to make their understanding explicit in many cases. Systems should encourage a flexible mixing of knowledge in the head with knowledge in the machine, of tacit and explicit, of intuitive and verbalized.
When it is deemed desirable to capture knowledge in a computer support system, then language can play a major role. According to the theory proposed here based on Heidegger’s analysis, the capture of knowledge is a refinement of the process of putting tacit preunderstanding into language. One way to make this process explicit and to help people exert control over it is to provide a language facility to support the expression of knowledge for computer capture. Because people need to relate explicit assertions to their tacit preunderstanding, it is important to tie the language facility to people’s natural modes of expression as much as possible. So, for instance, the computer-based language could use terms from the design domain chosen by the designers actually using the system. This means having most of the vocabulary user-defined and easy to extend or modify. The appearance of the language can be made similar to natural language also, to ease the translation back to the level of original discourse in which the intentional content is less alienated than in codified and formalized expressions of logical calculii. This can be accomplished to some extent by careful design of the syntactic appearance of the language facility, keeping this naturalness a priority.
Linguistic tools to support interpretation and collaboration in design should support the interplay of tacit and explicit understanding in the interpretation of the language. While it may be necessary to structure an end-user programming language in a way that can be used to control the computer like traditional programming languages, one wants to avoid the cognitive costs of using these languages as much as possible. Although it may be important to include some of the basic functionality of traditional programming concerns such as variables, recursion, types, control structures, etc., it is desirable to avoid requiring the users to keep the associated doctrine in mind. One wants a programming language in which most of these structures are kept tacitly behind the scenes most of the time.
At the same time, the user still needs to be able to analyze the structure of the representations explicitly and formally (e.g., as a hypermedia node and link structure) and have expressions in the language reflect that. For instance, if one wants to capture a rule of thumb involving the separation of public and private areas in a lunar habitat, then one must think through a scheme for operationalizing and representing the notion of privacy that is involved. This is an extension of the process of explication that is involved in all interpretation of innovations that are not adequately comprehended by situational preunderstanding. However, once the rule of thumb has been expressed (written) in the language, the explicit understanding should be able to resubmerge into a more tacit comprehension. That is, when the rule is used in the future—whether by the original creator or by a subsequent designer using the system and its language—it should be available (readable) in a more tacit way.The problem of tacit and explicit understanding pervades the design issues for a system of computer support of cooperative design. The design of a language facility for computer support must particularly address this problem because discourse is the medium through which one moves back and forth between tacit and explicit understanding within the process of interpretation.
Go to top of this page
Return to Gerry Stahl's Home Page
Send email to Gerry.Stahl@drexel.edu
This page last modified on January 05, 2004