A Pilgrim's Progress: From Cognitive Science to Cooperative Design
AI & Society, 4,4, Fall Issue, 1990, 259-275
Liam J. Bannon
Computer Science Department
Aarhus University, Aarhus C, DK 8000, Denmark
June 1990
Running Head: From Cognitive Science to Cooperative Design
A Pilgrim's Progress: From Cognitive Science to Cooperative Design
Abstract
This paper provides a glimpse of some different theoretical frameworks and empirical methods in the author's search for theories and practices that might improve the utility and usability of computer artifacts. The essay touches on some problematic aspects of currently accepted theories and techniques in the cognitive sciences, especially in their application to the field of human-computer interaction, and mentions some alternative conceptions based on a cultural-historical approach. The intent is to widen the nature of the debate about appropriate frameworks for discussing human activities especially when we discuss design activities in the context of computer systems development. The paper concludes with some suggestions for more fruitful research directions that involve the active participation of those for whom the research is ostensibly being done, and a greater emphasis on understanding how artifact design and use are inextricably intertwined.
Keywords
activity theory, cognitive science, cooperative design, human-computer interaction, participation
1. Introduction
For a number of years I have been interested in both the design and use of computer systems and in understanding human cognitive activities. This interest was prompted by my academic background in both psychology and computer science. There seemed to be interesting possibilities in combining aspects of these disciplines. One area of overlap is in artificial intelligence and the use of the computational metaphor in psychology, especially cognitive psychology. A second, more applied aspect focuses on how a better understanding of human psychology might be relevant in the design and use of computer systems - human factors aspects. I have been involved in both of these areas over the years. Whilst for a period these two topics were quite distinct, one theoretical, focusing on representation issues, and the other more applied, concerned with ergonomic issues, by the early 80´s attempts were being made to link the two in the field labelled "human-computer interaction". Cognitive psychology, and the new cognitive science (see below) wished to apply its theory to actual settings, and the design and use of computer systems seemed a rich domain, as well as having funding sources.
While I was happy to be a part of this emerging discipline, I had a nagging doubt about the relevance of much this research to actual work situations, and I certainly did not see very much evidence of how this work was making for a more human centered technology. My somewhat naive belief at that time was that work in these fields could be directly used to improve the quality of computer systems and work activities involving computing systems, as evaluated by those using them in their daily lives. Yet even a casual glance at many work settings involving computers today does not support this belief. Rather, I have come to believe, along with Rosenbrock (1981) and many others, that our present-day design and utilization of information technology in work has tended to restrict, rather than expand human potential. This lead me to question the underlying perspectives and values that lay behind certain approaches to systems design which I have attempted to surface - for example, the idea of “idiot-proof” systems (Bannon, 1986a). It also lead me to question the theoretical base for much of the cognitive ergonomics work, information processing psychology, for it did not seem to address issues that I saw as important. This questioning of the field and parallel search for alternative frameworks also made me more aware of the political aspects of work organization and how the socio-economic foundations of our society affect the nature and organization of all work activities, with or without technology. In other words, problems are due not solely to the nature of the technology itself, although it is not a neutral element (see Winner, 1980) but also to the organization of work around the technology (see Boddy & Buchanan, 1982), and the general socio-economic and political rationale within our society which develops these machines and industrial systems. In this paper I do not intend to single-handedly discuss all of these complex problems and their interdependencies, but I will try and demonstrate how a particular theoretical paradigm and associated empirical methods and procedures can blind us to many important issues regarding design and use of technology, and offer some suggestions for alternative ways of viewing and studying the design and use of artifacts that may improve the quality of work life for those involved with computer systems.
The title of this paper was chosen to indicate a personal quest for understanding people and computers that has taken me through some quite different territories, both literally , in terms of geographic localities, and metaphorically, in terms of ideas and disciplines, over the years - only some of which will be mentioned in this particular article. The sense of "pilgrim" which is intended here is therefore that of a person who wanders in foreign lands, rather than the more common spiritual connotation. In my case, the journey starts out from the disciplines of cognitive psychology and its close relation, cognitive science. After an encounter with these theoretical formulations and some critique, we advance to the field of human-computer interaction, and investigate what is being done and how it might be useful in the design of more "human-centered" systems. Problems again emerge and we continue in our search for alternative ways of improving work quality through artifact redesign that takes us to Northern climes and the discovery of such ideas as user involvement and understanding of the use situation, prevalent among many Scandinavian researchers. This essay finishes at this point, but in no sense is the journey itself finished, and some questions remain as we look further up the road. So, let us begin our journey in the field of cognitive science and investigate some of the problems inherent in its theoretical premises.
2. Theoretical Frameworks
2.1. A Brief Outline of Cognitive Science
Cognitive Science is a multidisciplinary activity, spanning a wide variety of fields. Branches of the standard disciplines of psychology, sociology, anthropology, computer science (especially artificial intelligence), linguistics, philosophy, neurophysiology and biology, can be considered as part of cognitive science, or the cognitive sciences, as some prefer to label the area. They supposedly share an interest in the origins, development and nature of intelligence in organisms of all kinds, even in machines. A key point is that one can separate out the actual implementation of the system, in biological or physical matter, from its functional description. The focus is on the search for mechanisms that could give rise to intelligence. One of the underlying assumptions shared by many in the field is that what underlies all intelligence is the ability to manipulate symbols. Newell and Simon (1976) put forward the empirical hypothesis, that a physical symbol system has the necessary and sufficient means for general intelligent action. Most cognitive scientists accept the essential similarity of processes that are behind human and artificial "reasoning". Pylyshyn (1984) is most explicit about this: "...my proposal amounts to a claim that cognition is a type of computation" (Preface, pg xiii). This view has lead many researchers in the field to build computer models of human thought processes that are taken to be strongly equivalent to what it is people do in their comprehension and understanding of the real world. Most of this work accepts what has been called the classical representational theory of the mind (see Fodor & Pylyshyn, 1988). This holds that there is a “language of thought”, that mental representations have a combinatorial syntax and semantics, that both brains and digital computers are physical symbol systems, etc. It has held sway for more than a quarter of a century, influencing theory and practice (in the sense of empirical studies) in cognitive psychology, AI, and branches of linguistics and philosophy. While it will not concern us here, it is interesting to note that within the cognitive science community the ascendancy of the "classical" account has recently been contested by those, often with background training as physicists, who argue for distributed neural network architectures or connectionist models. An example of this debate can be found in the debate between Smolensky (1988) on the connectionist side, and Fodor and Pylyshyn (1988) for example.
Despite the often exaggerated claims made by proponents of the cognitive science perspective as to its achievements, a number of people have admitted there are problems in this approach. Donald Norman, one of the pioneers in the field, as early as 1980 wrote a prescient paper that pointed to the gaps existing in the then fledgling new science (Norman, 1980). He was particularly concerned about the basic building block in this approach, which he refers to as the model human-information processor.. " The problem seemed to be in the lack of consideration of other aspects of human behavior, of interaction with other people and with the environment, of the influence of the history of the person, or even the culture, and of the lack of consideration of the special problems and issues confronting an animate organism that must survive as both an individual and as a species..."(pg.2) . Ten years later, much of this critique remains unanswered if we investigate the state of the field today. What has been changing, at least in some circles, is the uncritical acceptance of this paradigm as the only, or even a sufficient one to handle the queries he raises. While important issues were raised in the 1980 paper, the particular areas that Norman suggests for further research seem in many cases to ignore the specific concerns identified in the above quote. An alternative framework for considering the relation between organism and environment is not clearly stated, though there is a hint of such an alternative in the remarks by Cole that appear in an appendix to Norman's article.
2.2 An Brief Outline of Activity Theory
Recently, a number of challenges to the standard rationalistic mainstream cognitive science theoretical framework have been put forward. Some theorists focus on the individualistic nature of much cognitivist theorizing, arguing for greater attention to the setting in which cognition takes place and how it is shaped by this setting (Lave, 1988). Some have argued for a radically different epistemology for the discipline, eschewing the “Cartesian model” for a hermeneutical interpretation. Such a radical critique has been popularized within the computing fraternity by the work of Winograd and Flores (1987) starting out from the work of Heidegger, Maturana and others. Another long-standing critique of Cartesianism comes from the dialectical materialist tradition within Soviet thought, developed from ideas about activity present in the work of Hegel, then elaborated by Marx and applied within a psychological framework by psychologists such as Vygotsky (1978) and Leontiev (1978). This cultural-historical or activity approach is the one that I will mention here, as it has a number of interesting features within our present context.
Perhaps the term "Activity Theory" is somewhat of a misnomer for the perspective, for although the concept of activity is central to this approach, there is not some monolithic theoretical superstructure that is accepted as defining the theory. Although elaborated most completely in the domain of psychology, the concept is not restricted to this domain. The theory is difficult to comprehend without a background in German philosophy and and Soviet thought, and the reader is forewarned that the following comments barely scratch the surface of the theoretical concepts that underlie this perspective. For further accounts of this approach, see Engestrom (1987), Kozulin (1986), Kuutti (draft), Raeithel (in press), Wertsch (1981, 1985) and Leontiev (1978).
Kozulin (1986) has studied the development of the concept of activity in Soviet psychology and traces its origins back to the attempt to replace behaviorism with something more materialist and dialectical in its essence. He notes “According to Vygotsky, human behaviour and mind must be considered in terms of purposive and culturally meaningful actions rather than as biological adaptive reactions. Objects of human experience - and therefore objects in psychological experiments - are socially and culturally meaningful things and not just abstract stimuli. Activity then takes the place of the hyphen in the formula S-R [Stimulus-Response, the basic unit of Behaviorism], turning it into a formula, object<->activity<->subject, where both object and subject are historically and culturally specific." (Kozulin, 1986)
By analogy to physical (material) tools, Vygotsky viewed language and symbol systems as powerful psychological tools in the development of the higher mental functions. They are artificial and social in nature. In this approach cognition is fundamentally social, first appearing at the interpsychological level - between people, then being internalized to the intrapsychological level. By analysing the available psychological tools in different cultures, and at different periods in history, Vygotsky believed one could find evidence for differences in the “higher human functions” in particular societies.
In a concise introduction to the activity concept Kuutti (draft) defines activity as the fundamental kind of context within which human actions take meaning. An activity has the following properties (Kuutti):
"an activity has a motive and activities can be separated according to their motives. A motive tells why activity exists.
-an activity is a collective phenomenon.
-an activity has a subject, who understands the motive of the activity. This subject can be individual or collective. All participants do not necessarily understand the motive of the activity in which they participate (thus the origin of alienation).
-an activity exists in a material environment and transforms it.
-an activity is a historically developing phenomenon.
-contradictions are the force behind the development of an activity.
-an activity is realized through conscious and purposeful actions of participants."
Of interest is the emphasis on transformation of activities and on their cultural mediation, which requires that we understand the historical development of the activity and its specific cultural setting. It is for these reasons that the approach is also referred to as the cultural-historical (mediation) theory or approach.
Backhurst (1988) has referred to this alternative tradition as “communitarian” in distinction to the “individualistic” Cartesian tradition and notes three key aspects of this approach:
"1 ) Activity (social forms of material activity) explains the nature and origin of human consciousness. "We become human through labour" (Leontiev).
2) The higher mental functions are social in nature and origin - in other words, to use a phrase coined by Cole et al. (1978), mind is “in society".
3) The higher mental functions are internalised forms of social activity " (Backhurst, 1988).
The concept of activity, especially within psychological theorizing, has been elaborated by Leontiev (1978) among others. Leontiev distinguishes between activities, actions and operations corresponding to motives, goals and conditions. Although these can be viewed hierarchically, with activities consisting of actions which themselves are implemented by a series of concrete operations, which are determined by the actual material conditions, these divisions should not be seen statically. Most of the empirical work within this tradition has focused on the level of actions and operations, and has some similarities to the later work of the German action theorists such as Hacker and Volpert (see Frese & Sabrini, 1985), though there are some important differences.
What is of interest in this approach is a more theoretical framing of certain issues that are of concern to me (see below) and which are difficult to conceptualize within traditional information-processing accounts of human behaviour. For example, the problem of context, which has become more and more recognized as a crucial issue for useful theory and empirical work is built into the very basis of the theory, in terms of activities. Having now discovered two alternative accounts of how to conceptualize human behavior and activity, let us move on and see how these different frameworks have affected the kinds of empirical studies done.
3. Methods of Investigation
In recent years there has been a wide ranging critique of our "scientific" theories and methods, from both philosophical and more pragmatic concerns. Much of this critique - of objectivism in science, of "pure" knowledge, of the distinction between rigorous facts vs. everyday experiences etc. - has been grudgingly accepted as having some base, but the critique of the scientific method itself, and how experimentation is performed has not been scrutinized or debated to the same extent. Thus students still read textbooks about "The scientific method" and learn to carry out experiments on "experimental subjects" as if they were carrying out experiments on a sack of grain!
What we have to realize is that experimenter and experimented are both human beings, with wishes and desires, even in the experimental manipulations, and that the attempt to suppress this systematically reduces the validity of the "experiment " that is being conducted. Such criticisms are not simply cries for more "humanity" in experiments, but argue convincingly for a stance that sees "subjects" in experiments as genuine subjects, not, as is usually the case, as objects to be manipulated at the whim of the investigator. In their forward to a collection of papers on new paradigm research, Reason and Rowan (1981) articulate a long list of items which they believe are inappropriate in much current research. Besides the model of the person, already mentioned, they mention reductionism, deception, reification, over-dependence on measurement and testing, bigness, sampling, detachment, conservatism, contamination, and scientific fairy-tales as problems for traditional research. Not a bad beginning. My purpose here is not to explicate each of these failures here, but to agree with the authors that most of the problems on this list can be found in the studies regularly conducted in many standard studies of the human sciences.
Just looking at cognitive psychology, for instance, we find many studies in which people are given apparently "meaningless" tasks to perform while their reactions are noted by the experimenter and later analysed in some representational formalism which is usually not shown, never mind understood, by the objects of research, the subjects themselves. Often these results are then generalised to account for everyday human cognitive functioning without any external validity checks. Recently, a serious questioning of the relevance of these studies to everyday human activities of which these experiments were a purported idealization has begun. These studies tend to analyse the individual without reference to their community, or their history, performing on a task designed by the experimenter in an unfamiliar environment . The "problem" is defined and valued by the experimenter, not by the subject, who is then expected to perform in certain ways. The question of how "subjects" make sense of the game in which they are playing, trying to discover the "rules of the game" i.e. what the experimenter is after, is often not explicitly discussed in these studies. Performance is measured relative to a certain "ideal", rational model of problem-solving, and the deviations of subjects from rational action is noted. It is presumed that the fundamental mental mechanisms posited to underlie human behavior in such prescribed domains can later be extended, without major modification, to more real-world activities. So, for example, the assumption is that "problem-solving" is a generic cognitive activity which has a similar form across a wide variety of domains, from acting in a psychology experiment to everyday cognitive activities. It is also usually assumed, implicitly, that this activity is located "in the head" of the individual. In an influential book, Lave (1988) provides a strong critique of both the theoretical underpinnings of the mainstream approach and the experimental manipulations often associated with it, arguing that “Cognition observed in everyday practice is distributed -stretched over, not divided among- mind, body, activity and culturally organized settings (which include other actors)".
My purpose in this paper is not to develop these arguments as to how the accepted research paradigm is flawed, but rather to note that just as the pervading information processing conceptual framework has been critiqued, a similar critique is developing over the nature of the research methodologies commonly associated with this approach. As we shall see, this conceptual and empirical baggage has has often been imported into the research on human - computer interaction without question, limiting the utility and usability of many HCI studies that are grounded in these debatable assumptions. To the extent that it does not give an adequate account of how people think and act in the everyday world, basing the design of artifacts on such studies may not be the most fruitful approach to adopt. Let us now briefly outline developments in the human-computer interaction (HCI) area.
4. The Rise of the Human-Computer Interaction Field
Human -Computer Interaction (HCI) studies did not emerge from a vacuum. Since the early days of this century, there has been a field of human factors or ergonomics concerned with the fit between people and machines. Much of this early work focused on physical or physiological measures, work rate, fatigue, etc. Some of this work seems to eerily resemble that done by F. W. Taylor, whose methods and practices were codified into "scientific management" and earned him the hatred of labor. Certainly some of the criticisms of experimental procedure voiced above could be applied to this work, in terms of the view of human experimental subjects as objects, the lack of involvement of workers in the conceptualizations behind the experiments and their lack of involvement in the interpretation of the results, or even access to them. Notwithstanding these problems, it should also be noted here that early ergonomic work did result in improvements in the quality of working life, especially in the area of occupational health and safety, often through legislation.
With the advent of the computer a much wider range of people than traditional ergonomicists became interested in the problems experienced by "users" of computer systems. Major influences in this expansion were cognitive psychologists, who realised that making a good computer interface was a matter not simply of physical, but increasingly, of cognitive ergonomics, and of software engineers, who were experimenting with the design of highly interactive interfaces, becoming concerned about how to conduct dialogues with users and how to present complex information to users effectively. The need to provide more adequate interfaces to computer systems grew with the rapid growth of discretionary users, people who saw themselves as having a job or profession that was not primarily geared to the computing medium itself, but wished to use it as a useful tool. Computer companies realised how "user-friendly" interfaces could give significant competitive advantage and dramatically increase the market for their products.
How might the academic field of cognitive psychology contribute to better computer system design? In an illuminating paper, Landauer (1987b) provides a useful framework for discussing this issue which we will follow. Firstly, he notes that there might be stores of knowledge developed within cognitive psychology that could be directly used in the applied field. Unfortunately, this is rarely the case. Aside for some general findings, for instance regarding the relationship between movement time, accuracy, and size of target (Fitt´s Law) which has been useful in comparing performances of pointing devices such as mouse, joystick, step keys, etc. the body of findings that can be applied directly is minimal. Much of the problem here is that it is difficult to generalise findings from laboratory experiments to real world contexts. Too much of importance has been left out in the effort to make for a "clean" psychological study! So much for results, what about the theoretical frameworks we have developed in psychology? Can they give us insight into the nature of cognition and provide new insight into how to design better human -computer interfaces? Part of the problem is that the gap between theory and application is too large. It is not at all easy to see how to “ground” some of the more developed cognitive theoretical frameworks in system design issues. One model of the human as an information processing system that was developed specifically as an engineering, calculational model, which would be of use in a design context was that of Card, Moran, and Newell (1983). A model user could be constructed from a basic cognitive architecture - a model human information processor - together with a set of approximate measures of human performance based on earlier experimental data. In principle this allowed a designer to predict the time involved in performing various operations on a system, so quantifying the tradeoffs that previously had to be guessed. While of some use in certain limited contexts, this approach has not been shown to have the flexibility, coverage, and ease of use necessary to make it a really useful instrument for designers. Its utility is more obvious for description of processes than understanding how to do things differently - which is what real design should be about.
A third possibility would be to utilize the sophisticated methods and techniques used in psychology to analyse user behavior, so that designers could find out how users perform on different versions of a system, or different prototypes, and not have to rely on intuition. Standard lab experimentation is too limited and costly and time-consuming. What is needed are "quick and dirty" methods that can give rapid feedback to designers about the utility and usability of their products. We are only beginning to develop such instruments.
So it would seem that despite the claims made for the role of cognitive science in advancing our understanding of HCI, and despite quite large investments of time and effort, our results are somewhat meagre. Even within the Cognitive Science HCI world, there are voices raised in concern at the level of adequacy of our theoretical accounts, and the low level of generalizability of many experimental results. Landauer (1987a, 1987b) decries the poverty of many of our experimental manipulations, and attempts to push psychology out of the lab setting in order to be more directly relevant to human needs in the workplace. He notes: "There is no sense in which we can study cognition meaningfully divorced from the task contexts in which it finds itself in the world". Whiteside and Wixon (1987) give some nice examples of how far removed some of the cognitive science work is from real world situations. It is this lack of appreciation of the use setting that is a major problem with much of the cognitive science HCI work to date. The body of useful knowledge available for designers is minimal. Advances in HCI seem to emerge from work groups without any clear lineage from the conceptual frameworks or empirical methods touted by cognitive science.
On a personal note, my interest in HCI was focused on user support issues and I found the cognitive science paradigm of little use in investigating how people actually managed to learn and use computer systems. At this time, I was concerned with how to help users learn about computers, to accomplish tasks, possibly through interaction with other users, perhaps even mediated through the computer (see Bannon 1986b, 1986c). In a sense, it seemed that in certain cases, a preoccupation with abstract theory can blind one to everyday practices which can be important to understand how people learn. Controlled experiments on command names seemed very removed from what was needed. Yet unguided observation and then speculation is also not sufficient. What to do? It was at this stage that I began to formulate my criticisms of the HCI field and its theoretical underpinnings and search for some inspiration. First, let me summarize some of the the criticisms:
The human-information processing model of the human is too limited, and too fractionated, neglecting important aspects of human activity. As noted by Velichkovsky & Zinchenko, (1982) this "boxology" has isolated cognitive processes from actions ( we are back to Guthrie's comment on certain theories, that they leave the organism "buried in thought"...!). The social nature of human learning is downplayed. Questions of motivation in the performance of tasks are not considered sufficiently. The way people perform activities in everyday life is not studied adequately, with the assumption that lab studies can be generalised readily to many work situations. A focus on representational formalisms predominates. The intent is often to model human activities with a view to their eventual substitution by computing procedures, without a clear understanding of exactly what aspects of the work lend themsleves to this simple substitution model, and what aspects would be more amenable to a view of "augmenting" the worker with computer tools. The underlying model of the “user” seems at times patronising and misguided - naive users, idiot-proof system design, etc. The possible relationship between such underlying values and extant practices in the professional community will not be developed here, though I believe they exist.
Faced with these problems, I began to collate the problems existing within the HCI field and argue for a shift in our perspective in order to find ways in which our work might become more relevant to everyday situations. I develop this analysis in Bannon (in press) and so only present a brief vignette from that paper here. One problem is that the majority of HCI studies to date take as their focus the individual user working on a computer system. This is adequate for certain purposes, yet the uncritical acceptance of this situation as the norm in the field has meant that technical support for the ongoing conversations and work-related activities that span groups of people in real work situations have often not been handled properly. Workers often have difficulty in coordinating their activities through the computer system. The system then becomes a barrier rather than a facilitator for the co-ordination of work. Extending the focus of concern from the human-computer dyad to larger groups of people and machines engaged in collaborative tasks is an important area for research in the next period. The quick growth of this field, labelled Computer Support for Cooperative Work (CSCW), attests to its importance (see Bannon & Schmidt,1989, for an introduction to this area).
Much of the early research done in the HCI field was confined to rather small controlled experiments, with the presumption that the findings could be generalised to other settings. Examples of such studies were those done on command naming conventions (see Barnard & Grudin, 1988 for a review of this research). It has become increasingly apparent that such studies suffer from a variety of problems that limit their usefulness in any practical setting. Firstly, by the time these studies are done the technology may make the original concerns outdated. Also important contextual cues for the accomplishment of tasks were often omitted in this transfer from the real world to the laboratory, and so the results of the lab studies became difficult to apply elsewhere. Increasingly, attention is shifting to in situ studies, in an effort to "hold in" the complexity of the real world situations, and a variety of observational techniques are being employed to capture activities, especially video. We can see an increasing focus on the concept of usability among the research community -whether people can and do actually use the resulting systems designed for them. It is difficult to evaluate the usability of an artifact without investigating the situations of use of that artifact. From a design perspective, this means that we need to have a prototype or test system for users to experience in order to get information on the usability of the resulting system.
Yet another problem is the tendency to focus on novice users in HCI studies or user testing, generally, with a concomitant neglect of the growth of competence and the development of people´s qualifications within work settings. Studying use of a system for a hour, or even a few days, does not give much insight into possibilities for long term growth and development of skills, measured over months, if not years.
Faced with a theoretical framework and associated set of empirical methods that did not seem to address many of the identified issues in computer systems design and use, the need for an alternative conceptual framework for understanding the issues and problems grew. Can we apply the alternative activity theory framework mentioned earlier to the field of HCI? An impressive effort at doing just that has been undertaken by Susanne Bødker (1987) in her thesis work. Shorter accounts can be found in Bødker (1989) and Bannon and Bødker (in press) so it is not necessary to outline the framework here. However let me note some of the interesting features of the theory , at least from my perspective. Fundamental to this approach is the mediation of activity through tools or artifacts. As noted earlier, such tools may include symbolic signs, language, as well as physical instruments. Tools are developed by a community over time, they can be viewed as crystallized knowledge and they are changed through use. Activities are accomplished through actions, which are in turn implemented as a series of subconscious operations corresponding to the context in which the action is being performed. Through learning we transform conscious actions into operations. However, if conditions change, then the flow of operations can be broken and they can again re-appear consciously as actions. In normal use situations our handling of artifacts is done through operations, and is not conscious to us. Thus Bødker notes that an artifact works well in our activity if it allows us to focus our attention on the real object, and badly if it doesn't, so we should talk about human operation of a computer application rather than of human-computer interaction .
This approach focuses on the character of the operations performed and the conditions under which they are activated. Through design we will change operations and their conditions, and in order to understand this, we have to allow the user to try out the new artifact in the work process, as we cannot predict in advance what future operations that accomplish an action will be. Here we have a firmer basis for arguing why some form of envisioning or prototyping is required in design.The role of practice within groups in the theory makes it possible to deal with human-computer interaction not just concerning an individual user but focusing on groups who share a practice. Design of artifacts is a process in which we determine and create the conditions which turn an object into an artifact of use. The future use situation is the origin for design, and we design with this in mind. Use, as a process of learning, is a prerequisite to design. Through use, new needs arise, either as a result of changing conditions of work or as a recognition of problems with the present artifacts.To design an artifact means not only to design the object which can be used by human beings as artifacts in a specific kind of activity. As the use of artifacts is part of social activity, we design new conditions for collective activity, e.g. a new division of labour, and other new ways of coordination, control and communication. Design of educational support is important too, because the artifact is to be integrated into an existing practice. A good education can facilitate this change.
What this very short sketch has hopefully demonstrated is that the activity theoretical approach provides an interesting framework for discussing human cognition and action. In particular, I find that the approach allows one to discuss not just individuals but relationships between people and settings in accomplishing work activities. The focus is on what people do with computer systems, not what the interface is like. Emphasis is on the use situation, as we have noted. The idea of a collective practice is an important one when we think of HCI implementations. Likewise the realization of the historical co-determination of both work settings and artifacts affects how we think about (re)design. The framework seems to "hold in" some of the complexity of real work situations when we start to think about developing computer tools for people, something that is missing in the individualistic cognitive science accounts.
5. Towards Cooperative Design
Our interest is on how to design usable computer applications, and on what kinds of research might support this process of design. Certainly we can learn some things from the evaluation of existing systems and problems people have with them, but our main question is how do we design to meet user needs. Traditionally in information systems development, this is done at an early stage by systems analysts who analyze current work activities and future needs. Such requirements analyses form the basis from which a system is constructed. Over the years, problems with this approach have been noted, particularly by Scandinavian researchers in systems development (see Bjerknes, Ehn and Kyng, 1987, Ehn, 1988, Greenbaum & Kyng, in press). These critiques cover a wide range of issues, only some of which will be mentioned here. Exactly what can be described in these descriptions is one issue (Ehn & Kyng, 1984). The form of these descriptions is another. Workers are often asked to evaluate the descriptions made of their work processes by analysts, yet this is often unproductive, as the representational formalisms adopted are often obscure to the workers. This point raises a more general one that I believe is worth noting, concerning, if you will, the politics of representations. We can question many aspects of this process. Who makes the representation, who has access to it, what purpose does it have? In many cases rather than clarifying things, they simply obscure actual work processes in a cloud of abstractions that make little sense to the people whose work is supposedly being modelled. Worse, these abstractions are then utilized as the basis for building the information system, with the result that the inadequacy of these descriptions becomes clear to all in the failure of the resulting system. Experiences in the office information systems area bear this out. All too often, as one researcher/developer noted, we end up "automating a fiction" (Sheil, 1983).
Adequate input from users is often presumed to have been captured in the requirements analysis phase of a project, and task analyses done earlier by the system team. Over the years, it has been acknowledged that these are often inadequate, and the question has begun to be asked whether this is because of some problems in the way of doing the studies locally, or whether there is a fundamental problem with the very assumption that one can map out users needs and requirements successfully in advance through simple techniques of observation and interviewing.
An alternative approach pioneered by different groups in Scandinavia, and sometimes referred to under the label "Cooperative Design" is to work with users in the whole design process, accepting that needs analysis documents will inevitably be incomplete, and that even with the best of intentions, users find it difficult to articulate how they work, especially if asked to use some form of modelling notation with which to describe or interpret it. Alternative ways of allowing users to map their work processes and the contradictions they currently experience are required. Such methods as Future Workshops, role playing exercises, games can assist this process. ( See Greenbaum & Kyng, in press, for examples) Many now argue that users need to have the experience of being in the future use situation, or at least an approximation of it, in order to be able to give comments as to the advantages and disadvantages of the proposed system. So, some form of mock-up or prototype, needs to be built in order to let users know what the future use situation might be like. Such cooperative prototyping studies, as exemplified by Bødker and Grønbæk,1989, are representative of what has been termed the “Scandinavian model” of systems development, where users are seen as equal partners in the development of systems (see the papers in Bjerknes et al., 1987). This perspective takes the work process as primary and attempts to support workers through providing them with skill-enhancing computerized tools (see Ehn and Kyng, 1987). What is required are various methods for envisioning future work situations, so users can, in (hypothetical) use, discover potential problems and make suggestions as to how to re-design the planned system. This involves intense commitment on the part of both users and designers to acknowledge each others competencies and inadequacies, and to attempt to construct a mutual dialogue. This can be enhanced through working with prototypes of systems where they can try out future practices, rather than comment on abstracted and ill-conceived flow diagrams of work processes. Such an approach presents a radical critique to much current systems development methodologies, yet acknowledged problems in traditional methods have focused more attention on this alternative perspective that has been developed by groups in Scandinavia over the last 20 years.
From the viewpoint of human-computer interaction, this alternative approach would imply moving a step beyond a user-centered view to a user-involved view. The term "User Centered System Design" has been used for some time, in order to focus attention on the needs of the user rather than on the system hardware and software possibilities. Yet what this phrase really means, or how it can be achieved, is far from clear. A more radical departure from much current thinking within the mainstream HCI world is to look on users not simply as objects of study, but as active agents within the design process itself. User involvement is not simply required to increase the effectiveness of the resulting system, as after all, in the last analysis, it is the users who know most about their work processes, but also to develop a more democratic work situation, so that those who will be affected by change have an influence on the kind of changes that will be made.
It is important to note that this concept of “Cooperative Design” is just that, a concept that needs to be fleshed out. I don’t think anybody who uses the term thinks of it as a “Method” or “Methodology” as such, indeed that is the last thing that should happen to the term...rather it suggests an ideal towards which people are striving. Based on experiences, we are learning about some techniques that seem to assist in this process, but it is very important to realise that you can’t simply coerce users into “cooperating”, nor can you be assured that you are doing “cooperative design” simply because you use certain techniques such as prototyping, or Future Workshops, or whatever techniques seem to be used by people discussing this concept. This is not to say that certain techniques may not assist in the process, the point is that it would be wrong to consider that the essence of the concept resides in the tools that are used - that kind of operationalization of the term vitiates the values underlying the approach, which respects users skills and seeks to promote more democratization in all phases of design and work practice.
6. Conclusion
We have travelled through some quite different conceptual and empirical territories in this search for better ways to think about and actually design computer based systems for work activities. We have visited various disciplines and their associated practices, and found much that we could take with us, and some ideas and practices that we feel should now be discarded. The journey continues, there is no end in sight, but perhaps it might be worth summarizing a few of the things we have learned so far along the road.
The gap between theory and practice is large. There are many aspects of human work activity that are difficult to comprehend within the standard individualistic information-processing account. There is scope for extension of the existing information-processing framework to include "distributed cognition" and for investigation of alternative theoretical paradigms. Our research methods often do not allow us to see how people perform in practical situations. Generalizing from laboratory settings to work settings can be dangerous. The issue of people´s work motivation is often not discussed. There are dangers in modelling human tasks and work processes, at the level of objectivity and of inhibiting shared understanding. In a design context, we should work with those whose work practices will be changing, as design and use are intertwined. Good design requires an empathy with the work process itself. I believe that we need to accept a more wholistic, value-laden stance where we as experimentalists or designers are not removed from, but are ourselves a part of, the process. Designers and users must be prepared to acknowledge each others competencies and to work towards understanding each other's practices.
In my search for more insightful theoretical frameworks for understanding human-computer interaction I have found the activity theoretical perspective sheds some new, though still hazy light. I look forward to hearing of the results of groups that have begun to work with this theoretical framework specifically in the area of human-computer interaction and systems design more generally. There is a long way to go before we can ground some of the concepts adequately. In my search for more useful empirical methods, I have also discovered some new possibilities under the rubric of cooperative design - future workshops, cooperative prototyping, role playing - ways of helping designers and users to understand each others practices. Here too, there is much scope for further research and for case studies of specific design projects in which these techniques are tried out and evaluated, in a joint activity involving both researchers and those involved directly in the projects.
It is not at all clear at this stage as to whether we can envisage some future state where the differing conceptual approaches mentioned here can be fruitfully combined. For the moment the important thing is that there is a growing recognition of the need to explore different paths, and to visit new places, and it is as a small contribution to this "opening up" that this particular essay is offered. The question of how the knowledge gained from a number of these different "pilgrimages" can be assessed and integrated may be left to others, at another time and place.
References
Bannon, L. (1986b). Helping users help each other. In D. A. Norman & S. W. Draper (op. cit.)
Bannon, L. (1986c). Computer-mediated communication. In D. A. Norman & S. W. Draper (op. cit.)
Engeström, Y. (1987). Learning by Expanding. Helsinki: Orienta-Konsultit.
Greenbaum, J. & Kyng, M. (Eds.) (In press) Design at Work. (Hillsdale: Lawrence Erlbaum)
Landauer, T. (1987b). Relations between Cognitive Psychology and Computer Systems Design. In J. Carroll (Ed.) (1987), Interfacing Thought: Cognitive Aspects of Human-Computer Interaction Cambridge, MA: MIT Press.
Lave, J. (1988) Cognition in Practice, Cambridge: Cambridge University Press.
Norman, D. (1980). Twelve Issues for Cognitive Science. Cognitive Science, 4 (pp. 1-32).
Sheil, B. (1983) Coping with complexity. Office: Technology and People, 1.
Vygotsky, L. (1978), Mind in Society, Cambridge, MA: Harvard University Press.
Wertsch, J. V. (Ed.). (1981). The concept of activity in Soviet psychology. Armonk, NY: Sharpe.
Winner, L. (1980) Do artifacts have politics? Daedalus, 109, 121-136, 1980.