Information Design Institute
The WWW is basically an access phenomenon: it provides easily accessible information to users for different purposes. How to design information for the web is the challenge faced by cognitive ergonomics today. To help frame this challenge, two models are presented: a web interaction model and a web information design model. These capture what are considered to be central features of information design that need to be addressed in the field.
The web interaction model involves knowledge expression on the part of the information provider and knowledge building on the part of the user, both channeled through the person's cognitive structure. Information on the web is categorized as data, complex information, or process and characterized along dimensions of coherence and dynamics. These are the elements that impact knowledge interfacing.
The principal concern for web information design is the less intentional nature of the communication involved and the interest-driven nature of web navigation. Both present difficulties for the traditional task-oriented design model of ergonomics. The web information design model presented considers the what (information selection) and the how (representation) of information. It suggests focusing on three design spaces in terms of access, interest, and transaction structures. These are the filters that need design attention.
The two models provide groundwork for further discussion of the issues raised and will hopefully encourage further modeling that will help establish this emerging field of knowledge technology as a prime concern of cognitive ergonomics.
The WWW has become a vast and constantly growing knowledge source for all kinds of users in all kinds of situations. The informal and varied nature of access to this knowledge source gives it a distinct status that differentiates it from more traditional information systems such as instructional software or database systems. The web embodiment of cyberspace thus provides its own challenges in terms of cognitive ergonomics. It is these challenges and their context that are explored in this paper, along theoretical lines that propose a view of how people interface with knowledge captured in easily accessible but relatively unstructured hypermedia environments found in cyberspace.
Two models are developed below to frame the discussion of these issues. The first proposes a view of the web as an interface between user and knowledge and is thus a representational model of this interaction. The second is a prescriptive model that captures the considerations that enter into the design of information on the web. Models are abstract tools that assist in thinking through what is involved in a given area of investigation. Their usefulness comes from focusing on certain features of the situation (possibly at the expense of other features, however) and in explicating the elements of the model and their interrelations. As such, models involve theoretical choices and are thus very open to discussion and counter-proposals - an ideal challenge to participants in a conference such as this one!
The discussion below is punctuated with aphorisms that seek to capture
the essence of the elements of the model and the gist of the views expressed.
These too are tools to assist communication and discussion - readers should
use their provocative aphoristic nature to further elaborate the underlying
concepts and challenge the representations they vehicle. The web is seen
by many, despite all the hoopla surrounding it, as the means for the emergence
of a very new cognitive habitus (Bourdieu, 1992) - a new cognitive framework
through which the individual interacts with society and by extension with
knowledge, on the order of revolutions such as those associated with printing
or television. Or as Fillmore (1995) puts it nicely: "Hyperliteracy involves
taking a leap into a new way not only of reading but of learning and living
as well" (p. 38). That is why a truly open discussion of cognitive ergonomics,
despite its abstract nature, is so essential and potentially rewarding
for the field.
A theoretical model for web interaction
The web interaction model presented in Figure 1 (which builds on more topical views related to learning - Duchastel, 1996a) focuses on the relationships between knowledge and information. Knowledge is the result of learning by individuals and is thus highly personal, well structured, and generally functional (useful for something relevant to the individual). Knowledge is in the head and only there. It can be expressed (communicated publicly) and it can be constructed (through learning, via interactions with one's environment, including the informational environment - Wilson, 1996; Duffy & Jonassen, 1992).
Figure 1: Web Interface Model
Information is the go-between that links the provider and the learner
or user. It is expressed knowledge that has been formatted into some representation
for the sake of communication, that is, in order to make it useful (for
instance for learning, for decision-making, or for enjoyment). Information
is what populates the web. Indeed, web sites are simply information repositories
that are made easily accessible to eventual users.
Let's now delve into some specifics about knowledge and information,
starting with the latter. The model considers three types of information
and two information facets. Information is typed as data (D) when it is
arbitrary and simply given (a piece of information), whereas it is considered
complex information (CI) when it is structured into an account of something.
A third type is process (P) when the information becomes interactive and
strongly goal-oriented. Examples of these three types are:
|D||airline fares||stock market quotations|
|CI||multimedia information on Shubert's career||information on the nature of Zen|
|P||an online exercise to practice simple subtraction||an online tool to fill out an income tax form|
The two facets of information relate to dynamics and to coherence. Information
varies in being relatively dynamic (changing) or static - e.g. stock prices
are more dynamic than air fares, which in turn are less static than physics
facts or subtraction rules. The facet of coherence is less well known in
HCI but just as important. It relates to how well information elements
cohere together (as in an argued defense of a case, or in an explanation
of Darwinism) or by contrast how they show arbitrariness (as in the names
for things, or telephone numbers).
The reason these types and facets are emphasized in the model are because they impact knowledge expression and construction, and hence how we interface with knowledge. Knowledge itself has been categorized in a number of ways (cf. Ryle's, 1949, distinction between what and how to, which, under the terms declarative vs. procedural, has been much debated in artificial intelligence - Gardner, 1985 - and used in cognitive psychology - Anderson, 1995; cf. also Jonassen's distinction between structural knowledge and other types, such as reflective procedural knowledge or imagery - Jonassen and Henning, 1996). Even now, knowledge analysis remains a relatively ill-structured domain.
Despite these difficulties, it is generally recognized in cognitive science (Gardner, 1985) that cognition is heavily filtered by one's existing cognitive structures, including one's intentions in using the information accessed in the environment. [Incidentally, this view of cognition has been a foundational truth of Buddhist metaphysics for millennia now - Cleary, 1990, pp. 25-26.] The process of browsing the web and accessing information is thus far from objective, even after we take personal interests into account.
The provider side of the web is just as complex. Not only does the provider - whether a single author or an organized team - need to exteriorize its knowledge in an information design that will impact the viewer, but it does so in a framework that is more or less intentional. Intentionality (discussed below) refers here to one's motives with respect to users: is the idea of web publishing just to express one's interests and expertise, or is it to influence a decision, or is it to change the user (for instance by teaching a skill or an appreciation of a domain of knowledge)?
In summary, knowledge is used by providers to create information artifacts
within a given cognitive environment, which in turn is accessed by users
for their own purposes and through their own colored glasses. The challenges
of cognitive cyberspace lie not in the diversity of what is to be found,
but in the cognitive distance created by less intentional forms of information,
which however arises from the diversity. Let's consider some of these challenges.
Challenges for knowledge interfacing
While an interface can be defined as a design that facilitates access to a given functionality in an object or system, the non-intentionality of much of what is made available on the web provides the first challenge for knowledge interfacing. Even though specific interfaces (such as learning interfaces) are geared to achieving specific outcomes (such as learning outcomes), knowledge interfaces are generally less task-oriented and rely more on user curiosity as a driving mechanism, despite their underlying communicative intent of a broad nature. They are similar in this respect to expressive designed artifacts such as art, and indeed, many personal home pages, for instance, are often more expressive than intentional - they largely aim to entertain rather than to inform.
The creative side of web design either camouflages to some degree the knowledge intentionality of the information or bypasses it to some extent. This is not a problem as such, for the web is multi-functional, but it does pose a challenge in terms of interface design as facilitating some knowledge functionality. As an extreme analogy, it is somewhat as if our textbooks were combined with art books and that we had to wade through the resulting mix in order to find what we were after. Now of course, if information on the web was as cleanly categorized as textbooks and art books are in bookstores, the issue would be of little importance. But, given the web as web (with all this means for disorganization and venturesome innovation), that is not the case.
Diminished intentionality on the web leads to less of a link between
user and provider. The communicative exchange is more diffuse and the importance
of the artifact as artifact grows. Compared to a book, information on the
web is less of a direct interface between author and reader. Rather, the
cognitive distance between the two grows: on the one hand, the provider
creates an information artifact often for a very broad audience and a wide
range of needs, and on the other, the user accesses information in a context
that is directed by personal goals and interests but that may not be ideally
matched with the information offered. The cornerstone of ergonomics, namely
the functionality that underlies task analysis, is put in question.
Part of this state of affairs is due to the information-rich / process-poor quality of the web currently. As this changes in the near future, we may see a reaffirmation of intentionality on the web. That will always be alongside much less-intentional information, however, so that dealing with this challenge in design terms will remain an important issue for cognitive ergonomics.
Primacy of interest
A second crucial challenge for knowledge interfacing centers on the imperative of interest. The growing availability of knowledge on the web and the main style of access, browsing, focus attention on the two principal aspects of information design, content selection and content representation (the what and the how of information). Design issues that arise in this context must be considered in light of web navigation and how it is driven by interest.
Web navigation is the locus of interaction on the web. Two prototypical modes are present at this time: search and browsing. While search involves a rather focused, intentional. action on the part of the user, browsing is much more exploratory, and generally driven by the attractiveness of the links encountered during this exploration. In addition, browsing itself can be more or less restricted to a given personal interest (or open to a vast set of personal interests), similar in this respect to TV interaction, particularly in channel-switching mode.
A complicating factor arising in navigation is the emergence of agents. Already, search engines do simple agency for users when they are seeking specific information, while a-synchronous agents (e.g. nighttime searchers) open the door to more independent agents to which one provides direction within a freer context of search. This leads eventually to browsing agents whose mission is to retrieve interesting information, rather than well-specified information as in a classic search.
The advent of agents does not fundamentally transform the issue of knowledge
interfacing in cyberspace. Only, it can create additional cognitive distance
between provider and user, adding to the distance already alluded to previously
through mitigated intentionality. Agents of course are themselves an interface,
with their own design requirements, but that in itself is a separate issue
The challenge for knowledge interfacing is one of capturing, in terms that will become usable functionally (for search or browse), the parameters of interest of the individual user. The fact that these interests are to a large extent dynamic (Duchastel, 1996b) does not facilitate the task. The challenge is similar to the classic challenge of the librarian (pinpointing exactly what the user is after), although transposed into the realm of interests, a domain that is much more difficult to explicate than the purely semantic one.
Towards a design model for cyber ergonomics
Discussion of these issues leads to a model of knowledge interfacing that can underlie the design of information and of specific tools for web use. The model is outlined in Figure 2.
Figure 2: Information Design Model
Knowledge interfacing deals with how information is designed, that is,
how knowledge is transformed into information and made accessible to users
so as to become once again knowledge. A related, but different, design
task focuses on tools: tools for providers to assist them in information
design and tools for users to assist them in knowledge building. These
are practical tools, similar in spirit to UIMSs, whereas the model presented
here is a prescriptive but abstract model centered on information design
Information has two tightly interrelated facets that come into play in design: the what and the how of information. The what concerns information selection (what information to present and what to leave out), whereas the how concerns how the information is represented (questions of media and particularly structure). Both facets invoke the professionalism of the author-designer, for the field is far from having any hard and fast rules to orient the design process, as might be the case in more settled realms of ergonomics.
Content selection is perhaps particularly soft as a skill. Coverage
of a domain, in terms of wanting to be accurate and comprehensive, is indeed
antithetical to ease of use by the eventual user. Delicate trade-offs are
the norm, particularly when we accept the non-objective, constructive nature
of knowledge (Wilson, 1996). The complexity of the task arises from the
fact that it involves both a transformation of one's knowledge into information
(via one's own cognitive structures) and an expectation that the user's
cognitive structure can assimilate (or accommodate to) the information
as presented. Incidentally, that complexity is the reason why applied cognitive
science, such as instructional design, remains so difficult.
The how of information design, also largely creative (particularly when we consider the non-intentional aspect of the web), may however be a little easier to categorize. Three design spaces are involved ( the term design space reveals the inarticulateness of the model given the current state of the art): an access structure, an interest structure, and a transaction structure. These are psychological structures, even though they may be embodied in physical elements, in that they highlight considerations to facilitate access, interest and processes.
Access structure is a term arising out of text and graphical design (Waller, 1982) to denote those elements of information design that are overlaid on the information in order to enhance access to the sub-elements and hence increase flexible use of the information. In books, an index, headings, overviews and summaries are elements of such an access structure. Access structure elements function to assist the user in selectively focusing on the essential (what the author-provider considers to be most important and what the user believes to be the most interesting for him or her). Conceptual maps and other such structural representations (Jonassen et al., 1993) are particularly relevant to access structure.
Interest structure is similar in that it focuses attention on the need
to capture the user's attention and keep it if there is good fit with the
user's particular needs. Some elements of interest have begun to be formulated
in specific models, such as Malone's (1981) intrinsic motivation model
developed through an analysis of computer gaming (its categories are challenge,
fantasy, and curiosity). Even in the more constrained domain of learning
environment design, however, there is yet no well-defined technology for
interest structure (Duchastel, 1996b).
Transaction structure is a term borrowed from instructional design theory (Merrill et al., 1991), where it refers to the set of interactions between learner and instructional materials, particularly computer-based ones. Transaction is a well-chosen term because it connotes a formal exchange involving interdependencies. In an instructional setting, for instance, a practice question will invoke a response, which in turn will lead to feedback and an instructional decision regarding what to do next. While particularly relevant to instruction, transaction structure can be extended to all information exchanges involving processes. And it becomes particularly appropriate as process type knowledge starts to develop on the web.
The information design model presented here is necessarily abstract and merely provides a general guide to the elements that need to be considered in practice. The challenge for cognitive ergonomics is of course to articulate the sub-elements involved, rendering the model more precise and hence more open to criticism (and to improvement). That task largely lies ahead of us.
Knowledge interfacing is not a new challenge. Ever since the invention of writing have we been faced with the task of designing information resources to assist others in knowledge building. The web may not transform the nature of information production as much as it does so for information access, but it does configure the design situation in a different manner, such that different concerns come to the fore. It is some of these that I have attempted to point to in this discussion of interfacing in cyberspace.
The core conclusion arrived at is that the traditional functional basis
of ergonomics may not be adequate to the challenge. The user orientation
of contemporary HCI remains central, but the task analysis that underlies
current design models is too narrow for the less intentional, more interest-driven
nature of web interaction. A wider framework is needed, one that might
possibly be organized along design spaces such as access, interest and
Metaphors play an important role in how we view knowledge interfacing.
Considering web information as an abstract interface between a user and
the knowledge of others, with cognitive filtering operating at both ends,
may seemingly lead us to the choice of a communication metaphor for the
web, but the intentionality issue poses a problem with this choice. On
the other hand, considering the web as a broadcasting medium (another metaphoric
choice) does not do justice to its high interactivity. The cognitive distance
involved in the web (distance between author and user) lies somewhere in
between the two metaphors. It could be said that we are dealing with a
new medium in the area of knowledge interfacing, one whose exploration
is just beginning. We will gradually build a knowledge technology, not
one based on hardware or software visions, but rather a soft technology
emphasizing the design of information arising out of knowledge and oriented
A number of issues face us in unraveling the parameters of knowledge technology. Some relate to the sociology of the net, such as the emergence of agents, trolls, and avatars (and in more abstract terms, memes), all of which affect how the net is interacted with and hence with how one interacts with the knowledge environment surrounding oneself. Others are more conceptual and may have potentially more impact for knowledge technology. Two may be particularly significant because they affect our very relation with information and with knowledge.
The first is the shift out of the skin, so to speak (Anderson, 1995). Knowledge is our internal (in the skin) storehouse of models and rules that enable us to be effective in the world. However, as technology permits a greater and greater capturing of knowledge in tools in our everyday environment, we need less learning of certain kinds. Electronic performance support systems illustrate this trend very well, as do expert systems and even dumb hand calculators. How far this trend can be taken remains an open question (Donald, 1991).
The second conceptual issue is just as far-reaching. It concerns the
emergence of practical multimedia and its extension into virtual reality
and metaworlds. Much of our current knowledge is structured in mental models
that we represent in propositional form and that we generally exteriorize
in verbal form in structured text. But text (writing) is but a handy means
of communication which was invented in a far-away technology-poor context.
Current technologies enable us more and more to bypass text and deal directly
with the phenomena of interest to us, through audio and visual means and
soon through immersion in virtual representations of our world. What this
means for the fate of text is a very open issue.
Knowledge technology as a design form is both a challenge for cognitive ergonomists and a fascinating field for the more theoretical study of knowledge interfacing. Cyberspace is the setting in which we will undoubtedly confront the issues involved.
Anderson, J. R. (1995) Learning and Memory. An Integrated Approach. New York: Wiley.
Bourdieu, P. (1991) Réponses: Pour une Antropologie Réflexive. Paris: Seuil.
Cleary, J. (1990) A Tune Beyond the Clouds. Berkeley, Cal.: Asian Humanities Press.
Donald, Merlin (1991) Origins of the Modern Mind. Cambridge, MA: Harvard University Press.
Duchastel, P. (1996a) Learning Interfaces. In T. Liao (Ed.) Advanced Educational Technology: Research Issues and Future Potential. Berlin: Springer-Verlag.
Duchastel, P. (1996b) A motivational framework for web-based instruction. In B. Khan (Ed.) Web-Based Instruction. Englewood Cliffs, NJ.: Educational Technology Publications. In press.
Duffy, T. and Jonassen, D. (1992) (Eds.) Constructivism and the Technology of Instruction: A Conversation. Hillsdale, NJ: Erlbaum.
Fillmore, L. (1995) Literacy's last best hope. On the Internet. 1 (4), 37-42.
Jonassen, D., Beissner, K. & Yacci, M. (1993) Structural Knowledge: Techniques for Assessing, Conveying, and Acquiring Structural Knowledge. Hillsdale, NJ: Erlbaum.
Jonassen, D. and Henning, P. (1996) Mental models: Knowledge in the head and knowledge in the world. Proceedings of the Second International Conference on the Learning Sciences, Evanston/Chicago, USA, July.
Malone, T. (1981). Toward a theory of intrinsically-motivating instruction. Cognitive Science, 4, 333-369.
Merrill, D., Li, Z. and Jones, M. (1991). Instructional transaction theory: An introduction. Educational Technology, 31 (6), 7-12.
Ryle, G. (1949) The Concept of Mind. London: Hutchinson.
Waller, R. (1982) Text as diagram: Using typography to improve access and understanding. In D. Jonassen (Ed.) The Technology of Text. Englewood Cliffs, NJ: Educational Technology Publications.
Wilson, B. (Ed.) (1996) Constructivist Learning Environments: Case Studies
in Instructional Design. Englewood Cliffs, NJ: Educational Technology Publications.