Supporting Situated Interpretation
Gerry Stahl
Department of Computer Science,
School of Environmental Design,
and Institute of Cognitive Science
University of Colorado
Boulder, CO 80309-0430
*****@**.********.***
large amounts of data. However, the problem is how to
Abstract
capture and encode information relevant to novel future
tasks and how to present it to designers in formats that
This paper discusses the role of interpretation in
support their mode of work.
innovative design and proposes an approach to provid-
A framework for clarifying the respective roles for
ing computer support for interpretation in design.
computers and people in tasks like lunar habitat design
According to situated cognition theory, most of a
is suggested by the theory of situated cognition. Several
designer knowledge is normally tacit. Situated
s
influential recent books (e.g., Sch n, 1983; Winograd
interpretation is the process of explicating something
& Flores, 1986; Suchman, 1987; Ehn, 1988; Dreyfus,
that is tacitly understood, within its larger context.
1991) argue that human cognition is fundamentally
The centrality of interpretation to non-routine
different from computer manipulations of formal
design is demonstrated by: a review of the design
symbol systems. These differences imply that people
methodology of Alexander, Rittel, and Sch n; a
need to retain control of the processes of non-routine
protocol analysis of a lunar habitat design session; and
design, but that computers can provide valuable
a summary of Heidegger philosophy of interpretation.
s
computational, visualization, and external memory aids
These show that the designer articulation of tacit
s
for the designers and support interpretation by them.
knowledge takes place on the basis of an understanding
From the viewpoint of situated cognition, the
of the design situation, a focus from a particular
greatest impediment to computer support of innovative
perspective, and a shared language.
design is that designers make extensive use of tacit
As knowledge is made explicit through the
knowledge while computers can only use explicit
interpretive processes of design it can be captured for
representations of information. This paper discusses the
use in computer-based design support systems. A
role of tacit understanding in designing, in order to
prototype software system is described for representing
motivate an approach to computer support of design. It
design situations, interpretive perspectives, and domain
focuses on three themes: (a) the need to represent novel
terminology to support interpretation by designers.
design situations; (b) the importance of viewing
designs from multiple perspectives; and (c) the utility
of formulating tacit knowledge in explicit language.
The Need for Computer Support
The following sections discuss how these three
themes figure prominently in analyses of interpreta-
The volume of information available to people is
tion in design methodology and in a study of inter-
increasing rapidly. For many professionals this means
pretation in lunar habitat design. Following a
that the execution of their jobs requires taking into
discussion of the tacit basis of understanding, the
account far more information than they can possibly
philosophy of interpretation defines interpretation as
keep in mind. Consider the lunar habitat designers who
the articulation of tacit understanding. Then
serve as a key example in this paper. In working on
consequences for computer support for interpretation
their high-tech design tasks, they must take into
are drawn, and they are illustrated by the HERMES
account architectural knowledge, ergonomics, space
system, a prototype for supporting interpretation in the
science, NASA regulations, and lessons learned in past
illustrative task of lunar habitat design.
missions. Computers seem necessary to store these
interpretation, representations were developed for
Interpretation in Design Methodology
describing pivotal features of the design situation that
had not been included in the original specification;
The centrality of interpretation to design can be seen in
perspectives were evolved for looking at the task; and
seminal writings of design methodologists. The
terminology was defined for explicitly naming,
following summaries highlight the roles of appropriate
describing, and communicating shared understandings.
representations of the design situation, alternative
The designers felt that a careful balance of public
perspectives, and linguistic explications of tacit
and private space would be essential given the crew s
understanding within the processes of interpretation in
long-term isolation in the habitat. An early design
design.
sketch proposed private crew areas consisting of a bunk
Alexander (1964) pioneered the use of computers
above a workspace for each astronaut. Space constraints
for designing. He used them to compute diagrams or
argued against this. The traditional conception of
patterns that decomposed the structural dependencies of
private space as a place for one person to get away was
a given problem into relatively independent
made explicit and criticized as taking up too much
substructures. In this way, he developed explicit
room. As part of the interpretive designing process, this
interpretations for understanding a task based on an
concept was revised into a reinterpretation of privacy as
analysis of the unique design situation.
a gradient along the habitat from quiet sleep quarters to
For Rittel (1973), the heart of design is the
a public activity area. This notion of degrees of privacy
deliberation of issues from multiple perspectives.
permitted greater flexibility in designing.
Interpretation in design is an argumentative process in
In another interchange related to privacy, the
the course of which an image of the problem and of the
conventional American idea of a bathroom was
solution emerges gradually among the participants, as a
subjected to critical deliberation when it was realized
product of incessant judgment, subjected to critical
that the placement of the toilet and that of the shower
argument (p.162). Rittel idea of using computers to
s
were subject to different sets of constraints based on life
keep track of the various issues at stake and alternative
in the habitat. The tacit acceptance of the location of
positions on those issues led to the creation of issue-
the toilet and shower together was made explicit by
based information systems.
comparing it to alternative European perspectives. The
Sch n (1983) argues that designers constantly shift
revised conception permitting a separation of the toilet
perspectives on a problem by bringing various
from the shower facilitated a major design
professionally trained tacit skills to bear, such as visual
reorganization.
perception, graphical sketching, and vicarious
In these and other examples, the designers needed
simulation. By experimenting with tentative design
to revise their representations for understanding the
moves within the tacitly understood situation, the
design situation. They went from looking at privacy as
designer discovers consequences and makes aspects of
a matter of individual space to reinterpreting the whole
the structure of the problem explicit. Certain features of
interior space as a continuum of private to public areas.
the situation come into focus and can be named or
The conventional American notion of a bathroom
characterized in language. As focus subsequently
was compared with other cultural models and broken
shifts, what has been interpreted may slip back into an
down into separable functions that could relate
understanding that is once more tacit, but is now more
differently to habitat usage patterns. Various
developed.
perspectives were applied to the problem, suggesting
new possibilities and considerations. Through
discussion, the individual perspectives merged and
Interpretation in Lunar Habitat Design
novel solutions emerged.
In this interpretive process, previously tacit features
As part of an effort at developing computer support for
of the design became explicit by being named and
lunar habitat designers, thirty hours of design sessions
described in the language that developed. For instance,
were videotaped and analyzed (see Stahl, forthcoming).
the fact that quiet activities were being grouped toward
The specified task was to accommodate four astronauts
one end of the habitat design and interactive ones at the
for 45 days on the moon in a cylindrical module 23 feet
other became a topic of conversation at one point and
long and 14 feet wide.
the terminology of a privacy gradient was proposed to
A protocol analysis of the designers activities
clarify this emergent pattern.
shows that much of the design time consisted of
processes of interpretation, i.e., the explication of
previously tacit understanding. As part of this
sketch a line, or how to look at a drawing and see the
The Tacit Basis of Understanding
rough relationships of various spaces pictured there.
The architect understands what it is to be a designer, to
Situated cognition theory disputes the prevalent view
critique a drawing, to imagine being a person walking
that all human cognition is based on explicit mental
through the spaces of a floor plan.
representations such as goals and plans. Winograd and
Heidegger defines the situation as the architect s
Flores (1986) hold that experts do not need to have
context including the physical surroundings, the
formalized representations in order to act (p.99).
available tools, the circumstances surrounding the task
Although manipulation of such representations is often
at hand, and the architect own personal or
s
useful, there is a background of preunderstanding that
professional aims. The situation constitutes a network
cannot be fully formalized as explicit symbolic
of significance in terms of which each part of the
representations subject to rule-governed manipulation.
situation is already meaningful (see Stahl, 1975). That
This tacit preunderstanding underlies people ability to
s
is, the architect has tacit knowledge of the situation as a
understand representations when they do make use of
whole; if something becomes a focus for the architect, it
them. Suchman (1987) concurs that goals and plans are
is perceived as already understood and its meaning is
secondary phenomena in human behavior, usually
defined by its relations to the rest of the situation.
arising only after action has been initiated: when
To the architect, a rectangular arrangement of lines
situated action becomes in some way problematic, rules
on a piece of paper is not perceived as meaningless
and procedures are explicated for purposes of
lines, but, given the design situation, it is already
deliberation and the action, which is otherwise neither
understood as a bunk for astronauts. The bunk is
rule-based nor procedural, is then made accountable to
implicitly defined as such by the design task, the shared
them (p.54).
intentions of the design team, the other elements of the
Philosophers like Polanyi (1962), Searle (1980),
design, the sense of space conveyed by the design, and
and Dreyfus (1991) suggest a variety of reasons why
so on indefinitely. This network of significance is
tacit preunderstanding cannot be fully formalized as
background knowledge that allows the architect to
data for computation. First, it is too vast: background
think about features of the design, to make plans for
knowledge includes bodily skills and social practices
changes, and to discover problems or opportunities in
that result from immense histories of life experience
the evolving design. At any given moment, the
and that are generally transparent to us. Second, it must
background is already tacitly understood and does not
be tacit to function: we cannot formulate, understand,
need to be an object of rational thought manipulating
or use explicit knowledge except on the basis of
symbolic representations.
necessarily tacit preunderstandings.
At some point the architect might realize that the
This is not to denigrate conceptual reasoning and
bunk is too close to a source of potential noise, like the
rational planning. Rather, it is to point out that the
flushing of the toilet. The explicit concern about this
manipulation of formal representations alone cannot
physical adjacency arises and becomes something
provide a complete model of human understanding.
important against the background of relationships of
Rational thought is an advanced form of cognition that
the preunderstood situation. Whereas a commonsensi-
distinguishes humans. Accordingly, an evolutionary
cal view might claim that the bunk and toilet were
theorist of consciousness such as Donald (1991) traces
already present and therefore their adjacency was
the development of symbolic thought from earlier
always there by logical implication, Heidegger proposes
developmental stages of tacit knowing, showing how
a more complex reality in which things are ordinarily
these earlier levels persist in rational human thought as
hidden from explicit concern. In various ways, they can
the necessary foundation for advanced developments,
become uncovered and discovered, only to re-submerge
including language, writing, and computer usage.
soon into the background as our focus moves on.
The most thorough formulation of a philosophical
In this way, our knowledge of the world does not
foundation for situated cognition theory is given by
consist primarily in mental models that represent an
Heidegger (1927), the first to point out the role of tacit
objective reality. Rather, our understanding of things
preunderstanding and to elaborate its implications. For
presupposes a tacit preunderstanding of our situation.
Heidegger, we are always knowledgeably embedded in
Only as situated in our already interpreted world can
our world; things of concern in our situations are
we discover things and construct meaningful
already meaningful in general before we engage in
representations of them. Situated cognition is not a
cognitive activity. We know how to behave without
simplistic theory that claims our knowledge lies in our
having to think about it. For instance, an architect
physical environment like words on a sign post: it is a
designing a lunar habitat knows how to lift a pencil and
sophisticated philosophy of interpretation.
significance; (b) preview or expectations of things in
The Philosophy of Interpretation
the world as being structured in certain ways; and (c)
preconception, a language for expressing and
Human understanding develops through interpretive
communicating.
explication. According to Heidegger, interpretation
In other words, interpretation never starts from
provides the path from tacit, uncritical
scratch or from an arbitrary assignment of
preunderstandings to reflection, refinement, and
representations, but is an evolving of tentative
creativity. The structure of this process of interpretation
preunderstandings and anticipations. One necessarily
reflects the inextricable coupling of the interpreter with
starts with sets of prejudices that have been handed
the situation, i.e., of people with their worlds. Our
down historically; the interpretive process allows one to
situation is not reducible to our preunderstanding of it;
reflect upon these preunderstandings methodically and
it offers untold surprises, which may call for reflection,
to refine new meanings, perspectives, and
but which can only be discovered and comprehended
terminologies for understanding things more
thanks to our preunderstanding. Often, these surprise
appropriately.
occasions signal breakdowns in our skillful, transparent
behavior, although we can also make unexpected
discoveries in the situation through conversation,
Computer Support for Interpretation
exploration, natural events, and other occurrences.
A discovery breaks out of the preunderstood
situation because it violates or goes beyond the network The theory of situated cognition and the philosophy of
of tacit meanings that make up the preunderstanding of interpretation stress how different human
the situation. To understand what we have discovered, understanding is from computer manipulations of
we must explicitly interpret it as something, as having arbitrary symbols. These theories suggest the approach
a certain significance, as somehow fitting into the of augmenting (rather than automating) human
already understood background. Then it can merge into intelligence. According to this approach, software can
our comprehension of the meaningful situation and at best provide computer representations for people to
become part of the new background. Interpretation of interpret based on their tacit understanding of what is
something as something is always a reinterpretation of represented.
the situated context. Representations used in computer programs must be
For instance, the lunar habitat designers discovered carefully structured by human programmers who
problems in their early sketches that they interpreted as understand the task being handled thoroughly, because
issues of privacy. Although they had created the the computer itself simply follows the rules it has been
sketches themselves, they were completely surprised to given for manipulating symbols, with no notion of what
discover certain conflicts among the interactions of these symbols represent. People who understand the
adjacent components, like the bunks and the toilet. Of domain must codify their background knowledge into
course, the discoveries could only occur because of their software rules sufficiently to make the computer
understanding of the situation, represented in their algorithms generate results that will be judged correct
drawings. The designers paused in their sketching to when interpreted by people. Only if a domain can be
discuss the new issues. First they debated the matter strictly delimited and its associated knowledge
from various perspectives: experiences of previous exhaustively reduced to rules, can it be completely
space missions, cultural variations in bathroom designs, automated.
technical acoustical considerations. Then they Many tasks like lunar habitat design that call for
considered alternative conceptions of privacy, gradually computer support do not have such strictly delimited
developing a shared vocabulary that guided their domains with fully catalogued and formalized
revisions and became part of their interpretation of knowledge bases. These domains may require
their task. They reinterpreted their understanding of exploration of problems never before considered,
privacy and represented their new view as a privacy assumption of creative viewpoints, or formulation of
gradient. innovative concepts. Software to support designers in
These themes of representing the situation, such tasks should provide facilities for the creation of
changing perspectives, and using explicit language new representations and flexible modification of old
correspond to the three-fold structure of interpretation ones. As the discussion of Alexander emphasized, the
in Heidegger philosophy. He articulates the
s ability to develop appropriate representations
preconditions of interpretation as: (a) prepossession of dynamically is critical. Because they capture
the situation as a network of preunderstood understandings of the situation that evolve through
processes of interpretation, representations need to be additional effort. The software should reward its users
modifiable during the design process itself and cannot for increasing the computer knowledge base by
adequately be anticipated in advance or provided once performing useful tasks with the new information, like
and for all. providing documentation, communicating rationale,
The concept of an objective, coherent body of and facilitating reuse and modification of relevant
domain knowledge is misleading. As Rittel said, non- knowledge.
routine design is an argumentative process involving
the interplay of unlimited perspectives, reflecting
differing and potentially conflicting technical concerns, The HERMES System
personal idiosyncrasies, and political interests.
Software to support design should capture these In Greek mythology, Hermes supported human
alternative deliberations on important issues, as well as interpretation by providing the gift of spoken and
document specific solutions. Furthermore, because all written language and by delivering the messages of the
design knowledge may be relative to perspectives, the gods. A prototype software system named HERMES has
computer should be used to define a network of over- been designed to support the preconditions of
lapping perspectives by which to organize issues, interpretation (a) by representing the design
rationale, sketches, component parts, and terminology. construction situation for prepossession, (b) by provid-
As Sch n emphasized, interpretive design relies on ing alternative perspectives for preview, and (c) by
moving from tacit skills to explicit conceptualizations. including an end-user language for preconception.
Additionally, design work is inherently communicative It supports tacit knowing by encapsulating (a)
and increasingly collaborative, with high-tech designs mechanisms for analyzing design situations using
requiring successive teams of designers, implementors, interpretive critics (Fischer, et al., 1993), (b) alternative
and maintainers. Software to support collaborative sets of information organized in named perspectives
design should provide a language facility for designers (Stahl, 1993), and (c) hypermedia computations
to develop a formal vocabulary for expressing their expressed in language terms (Stahl, et al., 1992). In
ideas, for communicating them to future collaborators, each of these cases, the hidden complexities can be
and for formally representing them within computer- made explicit upon demand, so the designer can reflect
executable software. An end-user language is needed upon the information and modify (reinterpret) it.
that provides an extensible domain vocabulary, is
usable by non-programmers, and encourages reuse and HERMES is a knowledge-representation substrate for
modification of expressions. building computer-based design assistants (like that in
Heidegger analysis of interpretation suggests that
s Figure 1). It provides various media for designers to
most of the information that would be useful to build formal representations of design knowledge. The
designers may be made explicit at some moment of hypermedia network of knowledge corresponds to the
interpretation during designing. One strategy for
accumulating a useful knowledge base is to have the
software capture knowledge that becomes explicit while
the software is being used. As successive lunar habitats for bunks?
are designed on a system, issues and alternative
deliberations can accumulate in its issue base; new
perspectives can be defined containing their own
modifications of terminology and critic rules; the
language can be expanded to include more domain discussion of issue;
What are the design considerations for bunks?
vocabulary, conditional expressions, and query
formulations. In this way, potentially relevant
information is captured in formats useful for designers,
because it is a product of human interpretation.
This is an evolutionary, bootstrap approach, where
the software can not only support individual design
projects, but simultaneously facilitate the accumulation
of expertise and viewpoints in open-ended, exploratory Figure 1. A view of the HERMES design environment,
domains. This means that the software should make it showing (left to right) a dialogue for browsing, a view of
easy for designers to formalize their knowledge as it the issue base, a critic message, a construction area, and
becomes explicit, without requiring excessive a button for changing interpretive perspectives.
design situation. Nodes of the knowledge Acknowledgments
representation can be textual statements for the issue
base, CAD graphics for sketches, bitmap images to This paper is indebted to Gerhard Fischer, Ray McCall,
illustrate ideas, sound for audio commentary, or Kumiyo Nakakoji, Jonathan Ostwald, and Tamara
language expressions for critics and queries. Sumner, who helped to focus it, and to the HCC
HERMES supports the collaborative nature of design research group at the University of Colorado generally,
by multiple teams through its perspectives mechanism. whose ideas and systems it tries to ground theoretically
This allows users to organize knowledge in the system and extend practically.
into over-lapping collections. Drawings, definitions of
domain terms in the language, computations for critic
rules, and annotations in the issue base can be grouped References
together for a project, a technical specialty, an
individual, a team, or an historical version. Every Alexander C (1964) Notes on the Synthesis of Form.
action in HERMES takes place within some defined Cambridge: Harvard University Press.
perspective, which determines what versions of Donald M (1991) Origins of the Modern Mind: Three Stages
information are currently accessible. in the Evolution of Culture and Cognition. Cambridge:
Harvard University Press.
The HERMES language pervades the system,
Dreyfus H (1991) Being-in-the-World: A Commentary on
defining mechanisms for browsing, displaying, and
Heidegger's Being and Time, Division I. Cambridge: MIT
critiquing all information. This means that designers
Press.
can refine the representations, views, and expressions
Ehn P (1988) Work-Oriented Design of Computer Artifacts.
of all forms of domain knowledge in the system.
Stockholm: Arbetslivscentrum.
Vocabulary in the language is modifiable and every Fischer G, Nakakoji K, Ostwald J, Stahl G, Sumner T (1993)
expression can be encapsulated by a name. The syntax Embedding Computer-Based Critics in the Contexts of
is English-like, in an effort to make statements in the Design. In Proc. of InterCHI 93.
language easily interpretable. The language is Heidegger M (1927) Sein und Zeit. T bingen: Niemeyer.
declarative, so users need not be bothered with explicit Translation: Heidegger M (1962) Being and Time. New
sequential programming concerns. Combined with the York: Harper & Row.
Polanyi M (1962) Personal Knowledge. London: Routledge &
perspectives mechanism, the language permits
Kegan Paul.
designers to define and refine their own interpretations.
Rittel H, Webber M (1973) Dilemmas in a General Theory of
This allows the HERMES substrate to support multiple
Planning. Policy Science. 4: 155-169.
situated interpretations.
Searle J (1980) Minds, Brains, and Programs. The Behavioral
and Brain Sciences. 3.
Sch n D (1983) The Reflective Practitioner. New York: Basic
Conclusion Books.
Stahl G (1975) Hermeneutics and Social Theory: Interpreting
and Transforming Our World. Ph.D. Diss., Dept. of
The theory of situated cognition argues that only
Philosophy, Northwestern Univ.
people tacit preunderstanding can make data
s
Stahl G (1993) Supporting Interpretation in Design. Journal
meaningful in context. Neither people nor computers
of Architecture and Planning Research.
alone can take advantage of huge stores of data; such
Stahl G (forthcoming) Interpretation in Design: The Problem
information is valueless unless designers use it in their
of Tacit and Explicit Understanding in Computer Support
interpretations of design situations. The data handling of Cooperative Design. Ph.D. Diss., Dept. of Computer
capabilities of computers should be used to support the Science, Univ. of Colorado.
uniquely human ability to understand. The philosophy Stahl G, McCall R, Peper G (1992) Extending Hypermedia
of interpretation suggests that several aspects of human with an Inference Language: an Alternative to Rule-Based
Expert Systems. In Proc. of the IBM ITL Conference:
understanding and collaboration can be supported with
Expert Systems (October 19-21, 1992).
mechanisms like those in HERMES for refining
Suchman L (1987) Plans and Situated Actions: the Problem
representations of the design situation, for creating
of Human Machine Communication. Cambridge:
alternative perspectives on the task, and for sharing
Cambridge University Press.
linguistic expressions. Together, situated cognition
Winograd T, Flores F (1986) Understanding Computers and
theory and Heidegger philosophy of interpretation
s
Cognition: A New Foundation for Design. New York:
provide a theoretical framework for a principled Addison-Wesley.
approach to computer support for designers situated
interpretation in the information age.
Supporting Situated Interpretation
Gerry Stahl
Department of Computer Science,
School of Environmental Design,
and Institute of Cognitive Science
University of Colorado
Boulder, CO 80309-0430
*****@**.********.***
Abstract
This paper discusses the role of interpretation in innovative design and proposes an approach to providing
computer support for interpretation in design.
According to situated cognition theory, most of a designer knowledge is normally tacit. Situated interpretation
s
is the process of explicating something that is tacitly understood, within its larger context.
The centrality of interpretation to non-routine design is demonstrated by: a review of the design methodology of
Alexander, Rittel, and Sch n; a protocol analysis of a lunar habitat design session; and a summary of Heidegger s
philosophy of interpretation. These show that the designer articulation of tacit knowledge takes place on the basis
s
of an understanding of the design situation, a focus from a particular perspective, and a shared language.
As knowledge is made explicit through the interpretive processes of design it can be captured for use in
computer-based design support systems. A prototype software system is described for representing design
situations, interpretive perspectives, and domain terminology to support interpretation by designers.
For publication in the Proceedings of the:
Fifteenth Annual Meeting of the
COGNITIVE SCIENCE SOCIETY
A Multidisciplinary Conference on Cognition
June 18-21, 1993
University of Colorado at Boulder