Академический Документы
Профессиональный Документы
Культура Документы
An activity theoretical approach to the design of a support for planning usability studies
Erik Markensten
Masters thesis in cognitive science Department of Computer and Information Science Linkping University, Sweden
Spring, 1999
Abstract
This masters thesis investigates into the possibilities to redesign the activity of planning usability studies in a large telecommunications company. The aim was to increase the potential for the activity to occur and develop in a specific direction. The framework employed is based on the sociocultural approach to mind. It is argued that this better explains cognition than the traditional cognitive approach. Activity theory has been used as a research framework and its usefulness in the study has been evaluated. The approach was to identify a number of contradictions within the activity, and in relation to other activities, that hinders the development in the wanted direction. Data was collected in interviews, workshops and during observations. The analysis included a qualitative analysis using grounded theory as well as an activity theoretical analysis. Since many of the contradictions found require a cultural change it is outside the scope of this study. A temporary solution was designed as a tool that will support the creation and development of the activity in the wanted direction. The main conclusions are that the contradictions exist not within individuals minds and intentions but in organisational and cultural structures. If all contradictions would be resolved the activity of planning for usability would not exist, since it is a consequence of the contradictions within and between the activity systems. It is also pointed out that there is a need for a more precise description of how to perform an activity theoretical analysis and a new term for cognition.
Keywords: Usability planning, planning, activity theory, sociocultural, cultural-historical, mediation, cognitive science, HCI, usability.
Acknowledgements
This thesis is going to print this afternoon. Of all the feelings that rush through me at this particular moment one of the strongest is that of relief. Writing a masters thesis can, apparently, be a frustrating job and it is most certainly a relief to finally finish. Despite this, I am absolutely satisfied with the process of getting here. During these last eight months I have learned and experienced more than ever in my education and it has been both exciting and fun. Much of this is due to the supervision provided by three of the most inspiring people that I have met in my education: Henrik Artman, Martin Howard and Pr Carlshamre. It is unlikely that this thesis would be the same, had it not been for the support of my supervisors, family and friends. In particular, I have been inspired and enlightened by late discussions on Nationernas Hus, spammed and amused by the postings on [kontakt] and welcomed by the people in Zelab. To avoid leaving someone out I will not mention any names but I owe you all big thanks. Tack!
Microsoft Excel and Microsoft Project are trademarks of the Microsoft Cooperation.
Contents
INTRODUCTION PROJECT BACKGROUND ASSIGNMENT AND OBJECTIVE RESEARCH QUESTIONS AND AIM THESIS OUTLINE TARGET GROUP FOR THE THESIS THEORETICAL BACKGROUND THE COGNITIVE APPROACH HUMAN-COMPUTER INTERACTION COGNITIVE SCIENCE THE SOCIOCULTURAL APPROACH MEDIATION A NEW UNIT OF ANALYSIS COGNITIVE SCIENCE FROM A SOCIOCULTURAL PERSPECTIVE PROBLEMS WITH THE COGNITIVE APPROACH THE LIMITS OF DUALISM, MENTALISM AND INDIVIDUALISM THE LIMITS OF RATIONALISM THEORETICAL RESEARCH FRAMEWORK THE CONTEXTUAL APPROACHES DECIDING ON A THEORETICAL FRAMEWORK ACTIVITY THEORY THE ACTIVITY SYSTEM THE HIERARCHICAL STRUCTURE OF ACTIVITY INTERNALISATION AND EXTERNALISATION DEVELOPMENT AND HISTORY MEDIATION CONTRADICTIONS SUMMARY OF THEORETICAL RESEARCH FRAMEWORK THE ROLE OF ACTIVITY THEORY IN HCI CONTEXTUALISING HCI ACTIVITY THEORY AS FRAMEWORK FOR HCI THE EMPIRICAL STUDY DATA COLLECTION METHODOLOGY INTERVIEWS AND OBSERVATION Usability Special Interest Groups meetings The Delta course Project planning session FUTURE WORKSHOP THE DELPHI WORKSHOP DATA ANALYSIS METHODOLOGY VALIDITY GROUNDED THEORY ANALYSIS PROCEDURE Activity theoretical analysis 1 2 3 4 5 5 7 7 7 8 9 9 10 11 14 14 16 19 19 20 20 21 24 25 26 27 28 28 29 29 32 35 35 36 37 37 37 37 38 39 40 40 41 41
RESULTS AND ANALYSIS OF THE EMPIRICAL STUDY PLANNING FOR USABILITY MISSING OUT THE TECHNICAL COMMUNICATORS PRIORITISING AMONG REQUIREMENTS SEPARATION OF USABILITY ACTIVATING VERSUS PASSIVATING CONFLICTING PERSPECTIVES Conflicts between PROPS and Delta Conflicts between SDPM and Delta REFLECTIONS ON THE PROBLEMS SUMMARY OF RESULTS DESIGNING SUPPORT FOR THE ACTIVITY THE DESIGN PROCESS DESIGN METHOD HOW TO SUPPORT THE NEW ACTIVITY Supporting the development of the activity Forces operating on the design DESIGN PROPOSALS Expert systems Information systems CONVERGING THE DESIGN Expert systems Information systems Exploring one alternative User evaluation THE FINAL DESIGN PROPOSAL THE WEB SITE THE BROCHURE DESIGNING ACTIVITY: RESULTS, DISCUSSION AND CONCLUSIONS RESULTS DISCUSSION DESIGNING AND CREATING AN ACTIVITY FROM TECHNOLOGY TO HUMANS WHY DESIGN A TOOL? THE USE OF ACTIVITY THEORY Activity theory as a theoretical framework Activity theory in design ACCOUNTING FOR USABILITY AS QUALITY AND DESIGN WE NEED A NEW DEFINITION OF COGNITION FUTURE WORK SUMMARY AND CONCLUSIONS REFERENCES
42 42 44 45 46 47 49 49 51 52 53 57 57 58 58 59 59 59 60 61 62 62 63 64 65 65 66 68 71 71 73 73 74 76 77 77 78 79 79 80 80 82
Designing Activity
Chapter 1
INTRODUCTION
As technology becomes more and more mundane in an increasing number of areas of use the need to emphasise quality as usability gets increasingly important. Technology, new as old, is not only something for a limited number of enthusiasts but for everyone. In fact, in modern societies, it is practically impossible to avoid any use of technology if one wants to live a fairly normal life. Those of us who have accepted this technological invasion and try to adapt to it sometimes have a hard job getting by. Why should one have to hesitate how to use the newly installed automatic teller machine for the fifth time or how to reserve the book you finally found in the library? Technological artefacts come and go and they are more and more accepted in our everyday life. Mundane technologies, such as mobile telephones, microwave ovens and video recorders are already, technologically and functionally, advanced enough. When comparing products such as those many focus more on other values than functionality, such as esthetical appeal and ease of use. Although research in human-computer interaction (HCI) and usability has been going on for thirty years, producing a wealth of process models, methodologies and tools for working with usability the industry has been slow to follow. On the other hand many of those companies that have adopted the new concepts have succeeded well and practice in these companies has often been ahead of research. Unfortunately those are not many and there is still a profound lack of work to improve the quality of products by focusing on usability issues. The traditional view on technology as that of simply providing functions has been prevailing for a long time. For a long time everyone who needed a computer at home have had no choice but to furnish their nicely designed homes and offices with grey hulks. Although designed computers have appeared before it was not until the introduction of the Apple Imac in 1998 that the traditional grey hulk was seriously challenged with an alternative. Nobody complained about or questioned the hulk designs, however, because this is the way it is had always been. Technology has only been for enhancing productivity and efficiency, and to make available more functionality. Due to the many difficulties involved in introducing usability in a company there is a body of research on the issue. Often the introduction of usability is done from within the company, driven by a small group of individuals. The difficulties arise when the two cultures of usability and technocentrism meet and need to converge. In large companies there are often well established process models, methods and routines for how to manage work. Equally often these models, stemming from an
Introduction
engineering culture, are not compatible with how usability work is structured. The consequences are difficulties in working with usability in these companies. Since there are already established methods and routines, usability, once introduced, is often seen as an add-on to the development process. There are usability teams and usability activities, which are not the same as the development teams and the design activities. The exact role of usability in the organisation is often vague and unclear and this is apparent in the projects. Often even planned usability efforts are rarely explicitly included in the overall project plan (Mayhew, 1999). More often, however, usability studies are not planned at all. When searching the available research for publications on the issue of planning usability studies in development projects the result is almost nil. There is practically nothing written on the issue. Instead, you find writings about how to plan and prepare usability tests, lots of it. There is equally little written about the whole usability process, from planning to final test. One of the firsts to grasp the whole picture, including the planing activity, was Mayhew (1999) in her book on the usability engineering process. The wealth of methods available for working structurally with usability can of course, implicitly, be regarded as a way of planning usability studies. The point here is that planning in this case is implicit, a side effect of using the method to structure work. The thoroughfare of the usability engineering process by Mayhew is the only book I have found that explicitly deal with the planning process. Even she, however, does not include the planning process as a natural step in the usability engineering process, as the first step in conducting a usability study. This study will investigate into the activity of planning usability studies. The theoretical framework that will be used is that provided by activity theory. On a more general level the use of activity theory as a framework and tool will be evaluated. The activity theoretical perspective is adopted to contextualise the unit of analysis. The concept of the activity, as it is mentioned here, in the research question and throughout the report, is that of a unit of analysis rather than any activity as in normal language use. This will be explained more in detail in chapter three but it is mentioned for the reader to be attentive to this interpretation of the term.
Project background
This thesis is based on a study that was conducted during the spring and summer of 1999 in co-operation with the Systems Engineering Lab (hereafter Zelab) within Research and Innovation at Ericsson Radio Systems, a major developer of products for managing mobile radio communication. Ericsson was established in 1876. Today it is one of the largest companies in the world offering advanced communications solutions for mobile and fixed networks, as well as consumer products such as telephones. Although Ericsson as a whole is immensely large with over 100000 employees, representation in 140 countries and a number of departments in a largely hierarchical organisation, this thesis only reports from one department in Linkping, Sweden. The business idea of the department in Linkping is to develop software applications for mobile radio telecommunication. This is not, as it may seem, applications for mobile telephones, but systems for managing, administering and supervising mobile telephone networks. These systems are gigantic compared to an ordinary office application. The development teams are equally large, ranging up to
Designing Activity
several hundred people, and the development time is about five years for the largest products. The customers are for the most part mobile telecom operators. Ericsson is in general a typically technique-driven business. This is not strange since much of the operations of the company are concerned with hardware such as telephones, network components and the like. In the applications area there has, however, been an ongoing work for changing this technique-driven view on product development by introducing usability in the organisation. This started with the introduction of a tailored usability engineering method, the Delta Method, in 1992. The Delta project was formed as a joint effort between Ericsson Infocom Consultants AB and the University of Linkping. The aim of Ericsson Infocom was to extend their software development methods to address usability and, in particular, to put the competence of the technical communicators to better use in the design process. The outcome of the project was a commercially available handbook in usability engineering that includes the competence of the technical communicator (see Carlshamre, 1994; Rantzer, 1996 for a more thorough report of the procedure of introducing usability in Ericsson). As Ericsson Infocom Consultants was merged into Ericsson the employees of Infocom, who by then were used to the new approach, were scattered in different departments. One of the key persons behind the work of the Delta method started to work in Zelab with usability matters in 1996. Since practically no one had heard about usability before Zelab soon became the main actor concerning usability in Ericsson in Linkping. Although there were many successful measurements taken to introduce usability in the organisation the progress was slow. Zelab had noticed how many people had had a difficult time planning for usability, even though they had an understanding of usability work. People with experience of planning activities in projects were asked to plan a usability study, which they knew about in theory, but had never worked with in practice. Often this resulted in unrealistic plans or an abandoning of the whole usability part of the project. This was considered a problem since a good plan for the work to be done is very important in project work. On the one hand planning allows one to more effectively manage ones own work efforts. On the other hand, getting the usability project plan included in the overall project plan increases the likelihood that it will actually be executed. The latter is important when it comes to usability matters since usability resources are often the first to get cut down. Since many of those involved in usability work were unsure of how to make usability plans they often turned to the experts in Zelab for help. Apart from taking resources from Zelabs main objective, to research, this constituted a bottleneck in the expansion of usability in the organisation. If someone at Zelab had got to review or help with every second usability plan made this would surely hinder spreading the usability perspective, at the same time as people would not learn themselves to make a proper usability plan.
A decision was made to approach the planning problem presented above by developing a supporting tool for usability planning that would help people to easier conduct usability plans. The problem of how this support could be designed and implemented was set up as a thesis assignment. This design problem is included
Introduction
into the main objective of this thesis; how the usability planning activity can be supported.
Why is it that the activity of planning usability studies is perceived as difficult? How can the creation and development of this activity be supported by the introduction of a new tool, and how could this tool be designed?
The study aims at extending the focus beyond the user and the (lack of) tools supporting the activity. The focus will be on the activity of planning usability studies and the problems connected to this. By focusing on the activity of planning, rather than on the possibilities of a new supporting tool a number of questions arise. What is the structure of the activity and how does it differ between individuals? Are there any inherent contradictions in the activity? If so, what do they originate in and how can they be eliminated? Which resources are available to the individual in the activity of planning for usability in terms of available tools, social relations, possibilities set up by the present rules and the individuals role in the organisation? In stating my first research question I intend to ask a more fundamental question than how we can make people plan for usability, namely, why this is needed. People within Zelab claim in fact that planning usability studies is actually not a complicated task. The assumption is that people find it difficult to plan for usability, because of the difficulties involved in shifting from an engineering perspective to a usability perspective on system development, just because it involves another perspective. By giving these people some practical help with a supporting tool the step would not seem as large. While I believe this is correct I want to take the issue yet another step and examine the factors that make this shift in perspective so difficult. The underlying assumption behind stretching the focus this wide is that there might be other more deeply-rooted reasons to why people find it hard to plan for usability. In approaching the above questions I will employ a theoretical framework grounded in activity theory. The activity theory has in recent years been suggested as a theoretical framework for HCI (e.g. Nardi, 1996). While many articles report using the activity theory, in my opinion, very few actually employ its full capacity as a tool and framework for HCI and design. This thesis is also an attempt to actually use the activity theory, not only as a post-hoc framework, but also as a describing, analysing and interpreting framework for the study. This is the second major issue in the thesis, to explore the usefulness of activity theory not only as an overreaching framework, but also as a practical tool in design. I believe that using the activity theory will help extending the object of study and find designs and solutions that support real problems, are well integrated with the current work, and which had maybe not been thought of, had I not extended the object of study. The use of activity theory as a framework and tool is covered by the third research question:
3.
What practical and methodological support does the activity theory provide when doing HCI studies?
Designing Activity
The focus is thus on the activity of planning for usability. Although the assignment is to find a supporting tool the focus will be set wide enough to search for other solutions as well and to uncover eventual dimensions of the problem that cannot be resolved by simply introducing a new tool. These solutions could, for example, constitute modifications of organisational rules and structures.
Thesis outline
The research approach can be characterised as proceeding through three phases. First theoretical studies were conducted to gain an understanding of the activity theory and its role in HCI, and how and why one should plan usability studies. The latter turned out to be very difficult to find anything about since practically nothing is written on the subject. Secondly, an empirical study was conducted. The aim was to collect data material to be able to model the present and future activities of planning for usability, and to gain an understanding of their characteristics. A qualitative approach was taken as it was judged that this would better suit the nature of the research problem, an understanding of organisational and social practice, than a quantitative approach. As the exact scope of the study was not known in advance it can be said to be explorative in nature. Although I intended to cover enough issues to be able to answer the research question stipulated above, I was aware of that the time and resources available were scarce and that it might well be that I would not get a fully coherent and representative picture of the work and practices under study. Finally a design was made for the prototypical tool that would support the activity of planning for usability. This was based on the results of the empirical study together with the theoretical research.
This is a masters thesis in cognitive science at Linkping University. The thesis work should correspond to twenty weeks of full-time studies. The content is foremost aimed at the people within Zelab who set up the assignment, students of cognitive science or people with similar background. Of course, it may also interest people with other backgrounds who want to learn more about the issues covered in this thesis.
Introduction
Designing Activity
Chapter 2
THEORETICAL BACKGROUND
This chapter goes through the theoretical development of human-computer interaction and cognitive science as research disciplines within the cognitive approach. The aim is to present the theoretical concepts of this approach, particularly within the field of HCI, in order to suggest an alternative approach in the next chapter. The alternative approach is introduced in this chapter under the heading The sociocultural approach.
Human-computer interaction
HCI has its roots in general theory and software psychology, a discipline with solid basis in experimental psychology. The intention was from the beginning to accumulate empirical knowledge by controlled experiments. The scientific methodology would ensure that the knowledge was (statistically) true and applicable to other instances of HCI. The notion of usability was to arrange the interface to fit the human as well as possible. This notion of usability continued to exist in subsequent work on cognitive engineering, human factors and ergonomics (Ehn & Lwgren, 1997). As the name indicates, the human factors approach aimed at fitting machines to the skills of humans, both physical and mental. Bridging from Taylorism the focus was still on the machine and its possibilities. The human was factored in into the whole equation of how to increase productivity and efficiency (Bannon, 1991). As computing developed a separation between operating and programming developed, but the focus was still on the functionality of the software rather than on its ease of use. Applications were developed by programmers for programmers, who could handle complex interfaces without problem. As the personal computer started to get widespread, and the use of software extended beyond the developing and technical community, a demand of more usable systems was noticed. The field of
Theoretical background
human-computer interaction emerged in the early eighties as a response to these changing conditions. Researchers within cognitive science, which was an established discipline at this time, argued for a better cognitive coupling between the human and the computer, claiming that the human factors approach lacked in theoretical motivation. Research was conducted on human computer dialogues, information presentation and visualisation etc. This demanded a new approach, which was further fuelled by the limited impact that general theory and its followers had had on professional software development. The new perspective that emerged was grounded in classical engineering, it drew upon the methodological foundations of general theory but oriented itself towards a paradigmatic model of applied science and engineering. It was named usability engineering.
Usability engineering () amounts to specifying, quantitatively and in advance, what characteristics and in what amounts the final product to be engineered is to have. This process is followed by actually building the product and demonstrating that it does indeed have the planned-for characteristics. Ehn & Lwgren, 1997, p. 301
Thus it is crucial to be able to set up measurable usability goals. Without such goals it is not possible to measure and verify if you have reached the designated quality level. Measurable goals also address the problem of when to stop iterating. Examples of measurable aspects of usability include user performance on specified tasks, flexibility of the design, learnability and subjective preference or degree of satisfaction. One surprising trait of traditional usability engineering is that the usability of a system is divorced from its utility. It is as if usability engineering is primarily concerned with the users efficient and error-free access to the services of the system, but not with the appropriateness of the services (Lwgren, 1995; Ehn & Lwgren, 1997). I will stop my review of human-computer interaction here, well aware of that it is incomplete. Usability engineering, for example, is today far from that I have described above. Many other interesting sub disciplines of HCI, such as contextual design and interaction design, have emerged during the last decade and the concept of usability has matured. For a review of the development of the usability concept see Ehn & Lwgren (1997). For a review of the development of HCI and the different directions and disciplines within see Lwgren (1995) and Preece (1994).
Much of contemporary thinking about the mind derives from Ren Descartes' distinction between the body and the soul and the roots of cognitive science go as far back as those of philosophy. Among philosophers, as well as among psychologist at the turn of the century, the favourite method of gaining insights into the nature of the mind was introspection. The inherent subjectivity of introspection led to severe methodological problems that led to the behaviourist revolution in the beginning of the 19th century. According to the behaviourist tradition, scientific reasoning should be founded on observational data that should be obtained from experiments. From these data knowledge could only be expanded by using logically valid inferences. As described by Gardner (1985), in 1936 Alan Turing proposed a hypothetical machine, later to be named the Turing Machine, which essentially formed the basis of the computer. This machine was in principle able to execute any given function and hence, in line with Churchs proof that everything that can be computed can be computed with recursive functions, compute everything that was computable.
Cognitive science
Designing Activity
Turings reasoning was further developed by Newell and Simon in their definition of the physical symbol system hypothesis (PSSH). According to this, a physical symbol system is necessary and sufficient to carry out intelligent actions. Such a system consists of a control, a memory, a set of operations and input and output. The job of such a system should be to manipulate formal symbols and, by concatenation, produce other symbols. The physical symbol system was adopted as the general architecture of cognition. Turing further suggested that one could program a Turing machine so that it would be impossible to distinguish its answers to an interlocutor from those contrived by a living human being (Gardner, 1985). This was called the Turing test. When the first machines appeared the exiting question of can it actually think and to what extent does it think like a human arose. McCulloch and Pitts interpretations of the firings of the neurones in the brain as sequences of zeros and ones, in analogy with the binary digits of the computers further fuelled the interest. The most important metaphor of cognitive science, the brain as a huge computer, was born. All these events, together with many others in psychology, neurology and linguistics marked the birth of cognitive science in 1956. Cognitive science is a multidisciplinary research tradition including in particular the disciplines philosophy, psychology, artificial intelligence, linguistics, anthropology and neuroscience. Gardner (1985) define cognitive science as:
a contemporary, empirically based effort to answer long standing epistemological questions particularly those concerned with the nature of knowledge, its components, its sources, its development, and its deployment. Gardner, 1985, p. 6
Cognitive science, as described by Gardner (1985) has two central beliefs. First, when speaking about human cognitive activities, it is necessary to speak about mental representations and to posit a level of analysis wholly separate from the biological and the neurological, on the one hand, and the sociological or cultural on the other. Secondly, the computer serves as the most viable model of how the human mind functions. Thus, cognitive science explicitly decided to leave out affect, context, culture and history as influences and dimensions of human cognition in the belief that these could be added on later on. The reasoning went that if one included these phenomena as well it would make the task to difficult and, in an effort to explain everything, one ends up explaining nothing.
Mediation
The most fundamental concept in the sociocultural approach is that of mediation, This concept is most explored by Wertsch (1998) in his book Mediated Action. Imagine a man screwing a screw with a screwdriver. Who is doing the action, the
Theoretical background
10
man or the screwdriver? It is a ridiculous question to ask but it does emphasis the central theme in mediated action. Obviously the man is not doing the job himself but neither does the screwdriver magically perform it without intervention. The movement of the hand means nothing without the screwdriver. The mans action is to screw the screw. The screwdriver mediates this action, or, put in another way, the screwdriver is the mediational means by which the man carries out the action. The cognition is bound to the activity, the tools and the action. The important thing to grasp here is that there exists an irreducible tension between the agent and the mediational means. Studies of either the agent or the mediational means are interesting and relevant only when they contribute to the understanding of how the elements combine to produce the mediated action under consideration. Wertsch (1998) gives another example of mediated action, this time involving an internal tool. Say that you were asked to multiply 343 by 822. You could probably come up with the answer 281946. If asked how you arrived at this conclusion you would probably reply I simply multiplied 343 by 822., maybe showing a calculation as in Figure 1. If you, in a second try had not been allowed to use any tools (such as paper, pencil or calculator) you would probably not have made it. Returning to the issue of mediation, was it really you who performed the action in the firsts case? After all you said I just multiplied. If it was only you who did it, why then did you have such difficulties in the second case? The answer is, again, that specific mediational means was involved that mediated the action of calculation. The spatial organisation or syntax of the numbers is an essential part of a cultural tool without which we have problems solving this problem. A major point here is that the tool is, in some sense, doing some of the thinking involved. This argumentation is applicable to the first example with the screwdriver as well, although, in that case, it is an external action that is distributed between the tool and the agent, rather than an internal. Hence, it is Figure 1: A calculation important to be aware of the need to set the whole functional system as the unit of analysis, in order to focus on mediated action and not just individuals and artefacts in isolation. The term functional system refers to the system set up by people and their mediating artefacts. This is not easy, according to Wertsch, since it is embedded in our culture:
when asking about someones ability level, we are usually asking about his or her skills in operating a particular cultural tool. It is crucial to address this question since agents may demonstrate outstanding skills when operating with one cultural tool, but only average skills when operating with others. This point is easy to overlook because the cultural tool to be employed in any particular assessment is usually a fixed and unquestioned part of the particular sociocultural context. Wertsch, 1991, p. 57
While cognitive science has approached individual processes in order to understand sociocultural phenomena other disciplines have begun with an account of societal phenomena and then, from this, generated accounts of individual, mental functioning. According to Wertsch (1998) this academic dispute over whether to
Designing Activity
11
give psychological or cultural or sociological processes analytic primacy reflects an endless and fruitless antimony between society and individual that is not rationally resolvable. He believes that the tendency to formulate the issues in either/or alternatives is counterproductive and misleading and has its root in how the opposing terms are understood. To avoid confusion Wertsch suggests that it is necessary to live in the middle of different analytic perspectives, i.e. to link, but not reduce, different perspectives to each other. A way to do this is to take human action and, more specifically, mediated action as the unit of analysis:
action is not carried out either by the individual or by society, although there are individual and societal moments to any action. For related reasons an account of action cannot be derived from the study of mental functioning or sociocultural setting in isolation. Instead, action provides a context within which the individual and society (as well as mental functioning and sociocultural context) are understood as interrelated moments. Wertsch, 1998, p. 34
Action can be internal (e.g. thinking) or external (e.g. doing), and it may be carried out by individuals or groups. The fact that the concept of action is not inherently tied solely to individual or social processes means that it is not limited by the society/individual dichotomy. According to Wertsch, much of the contemporary research in human sciences in general, and in psychology in particular, is too narrowly focused on the individual in isolation which is a severely limited stance. An important way to avoid this individualistic reductionism, he argues, is to recognise the essential role played by the mediational means or cultural tools in shaping human action. Mediated action provides a kind of natural link between action, including mental action, on the one hand, and the cultural, institutional and historical contexts in which such action occurs, on the other.(Wertsch 1998) One of the results of accepting this new middle perspective is that the boundaries between agent and action begin to erode. Instead of assuming that an agent, considered in isolation, is responsible for action, the appropriate designation may be something like individual-operating-with-mediational-means.
In this section I summarise and comment on the last chapter of Hutchins (1995) pioneering book Cognition in the wild in which he introduces a sociocultural approach to cognitive science. Unless otherwise stated the rest of this passage refers to that chapter. Hutchins, taking a sociocultural stance, describes the development of cognitive science and the possible reasons that made it separate mental activity from action, history and culture. He starts off citing Dennett about the context of Alan Turings discoveries: What do I do, he must have asked himself, when I perform a computation? Well, first I ask myself which rule applies, and then I apply the rule, and then I write down the result, and then I look at the result, and then I ask myself what to do next, and. Hence, originally the model cognitive system (Turing doing computation) was a person actually doing the manipulation of the symbols with his hands and eyes. He was visually and manually interacting with a material world. The important thing to notice here is that the cognitive properties of the human in this case are different than the cognitive properties of the system that is made up of the human in interaction with these symbols. The human with his/her mediational means is a sociocultural system.
Theoretical background
12
Now, for Turing the essentials involved patterns of manipulations of symbols. The whole point with the Physical Symbol System architecture of cognition (see the section Cognitive science) is that formal manipulation of symbols can be implemented in any hardware. But these patterns of manipulations of symbols expressly do not involve the psychological processes which a mathematician uses in order to accomplish manipulations. The essentials of the abstract manipulation of symbols are precisely not what the person does. What Turing modelled was the computational properties of a sociocultural system. The rulebook is replaced by abstract rules, also inside the computer. The mathematician who was a person interacting with a material world is neither modelled by this system nor replaced in it by something else. The only thing that remained of the person was the insideoutside boundary. This boundary was assumed to be the boundary of the skin and the skull. In fact it was the boundary of the formal system. Hence this is not a model of the processes engaged in by a person doing the task. These programs produce the properties of a sociocultural system rather than a person and this is what cognitive science failed to notice. In the discipline of artificial intelligence (AI), then, the brain was removed and replaced by a computer. The side effects were that the hands, eyes, ears the nose, and the emotions fell off in this replacement. The whole idea in AI was to create the computer in the image of the man by joining forces in information systems and psychology. Hutchins summarises this as:
The computer was not made in the image of the person. The computer was made in the image of the formal manipulations of abstract symbols. And the last 30 years of cognitive science can be seen as attempts to remake the person in the image of the computer. Hutchins, 1995, p. 363
Unfortunately, as follows from the discussion, the computer was not made in the image of the man but in the image of a sociocultural system. Ever since the computational metaphor for mind was introduced cognitive science has tried to remake the human in the image of the computer. The language we use for mental events is actually the language we should have used for sociocultural systems to begin with. Thus, speaking of a sociocultural system (humans and artefacts together) as positing cognitive properties and using language for mental events when speaking about this cultural system may seem odd, but is really just something that should have been done from the very start. The problem is that the computer metaphor for mind is so deeply rooted that it is very difficult to avoid using it and fall back to GOFAI (Good Old Fashion Artificial Intelligence) reasoning. As will be discussed in the next chapter this kind of thinking is embedded in our language, which make it even harder to discuss the issues. This is because language is a tool, just as any other (though more powerful), and it constrains and frames actions that are mediated by it, such as reasoning. The mistakes of AI to take a virtual machine enacted in the interactions of real persons with a material world and make it the architecture of cognition has consequences. Note that in the quote from Dennett above the symbols that Turing manipulated were originally outside the person on a piece of paper. When the brain was replaced by the computer the apparatus that fell of (hands, eyes etc.) was exactly the apparatus that was needed to interact with these symbols. Instead the symbols were put inside. Then the physical aspects were rendered irrelevant as constituting an implementation detail. It should now be clear that the enterprises in cognitive science do not model real cognitive systems. From the beginning, the idea was to defer consideration of affect, culture, context and history, until such time that there was a good model of how an individual
Designing Activity
13
worked in isolation. These things could be added on later. However, as I have tried to show here by reporting Hutchins work, this is the wrong approach to take:
Within this underlying theory of cognition there can be no integration of emotion, because the part of the cultural system that is the basis of the physical symbol system excludes emotion. The integration of cognition with action will remain difficult because the central hypothesis separates cognition and action by definition. History and context and culture will always be seen as add-ons to the system, rather than as integral parts of the cognitive process, because they are by definition outside the boundaries of the cognitive system. Hutchins, 1995, p. 368
Norman (1993) drives a similar argumentation claiming that the focus in cognitive science is not determined by what is really important but by what can be measured by todays tools:
Humans are extremely complex, the most complex entity ever studied. Each of our actions is the result of multiple interactions, of a lifetime of experiences and knowledge, and of subtle social relationships. The measurement tools of science try to strip away the complexities, studying a single variable at a time. But most of what is of value to human life results from the interaction of the parts: when we measure simple, single variables, we miss the point. Norman, 1993, p. 15
While cognitive scientists, ranging from psychologists to anthropologists, have spoken of culture as a collection of things Hutchins argues that culture is a process. In fact, he writes, it is a human cognitive process that takes place both inside and outside the minds of people. As described above Hutchins proposes an integrated view of human cognition in which a major component of culture is a cognitive process and cognition is a cultural process. The powers and limitations of the mind cannot be established without reference to culture. Taking a sociocultural stance involves radically reinterpreting cognition and other central concepts within cognitive science. The fact that cognition is a cultural process rather than mental and individual does not imply that one should stop speaking about internal representations. There is nothing wrong with studying internal mentalistic functioning of human beings and neither is it wrong to do this in laboratories rather than in the wild. What is wrong is to interpret the results in a too general fashion. Hutchins, in responses to the review of his book explains it elegantly:
It certainly seemed necessary to examine cognition in the wild in order to see the cultural nature of cognition. This is not because laboratory cognition is acultural, but because the tradition that pursues cognition in the laboratory has an investment in believing that the effects of culture can be controlled. Culture is not absent from the laboratory, but the setting is constructed in a way that directs attention away from it. So, while I do endorse a call for more studies of cognition in the wild, I want to be sure we are clear on why that is a good thing to do. The wrong reason to study cognition in the wild is the belief that experiments are a special setting, or somehow different in cognitive or cultural terms from other real-world settings. The reorientation I advocate is not framed in terms of the settings of cognitive performances (lab versus real world) but in terms of the stance we take with respect to the interpretation of the observed behaviour. As Lave, Suchman and others have pointed out, laboratory experiments are just another socially organized context for performance. In a recent special issue of the journal Cognitive Science, a number of authors (mainly opponents of the situated action view) mistake situated action for action that takes place in real world (i.e. non-laboratory) settings. In fact, cognition in the laboratory is just as situated as any other instance of cognition. The implications of the distributed cognition view is not that laboratory research should be abandoned in favour of 'real-world' settings, but that the way that behaviour that occurs in laboratory settings is interpreted should be changed to reflect the ways that subjects make use of cultural resources in the production of that behaviour. Once this is done, however, some of the putative advantages of the laboratory disappear, and perhaps laboratory research becomes less interesting because it will seem to provide fewer answers than was assumed. To reiterate, we need to look in
Theoretical background
the wild, not because that is where the real cognition is, but because that is a place where it is easier to see the cultural nature of cognition." Hutchins, 1996, p. 66f
14
Hutchins demonstrates in his book that one can be in the wild and yet better equipped empirically than in any psychological laboratory with all the variables controlled. The amusing paradox, as Latour (1996) puts it, is that in leaving psychological laboratories for the deck of a helicopter carrier, Hutchins has managed to build a better laboratory, equipped with video and tape recorders and was able to describe in excruciating details the achievements of navigational tasks.
The common denominator for these three -isms is related to context. All three strive to find ways to ignore context when studying cognition and all three have influenced the cognitive paradigm. I will discuss them here and argue why they should not have been taken as grounds for an approach to study and understand cognition. Cognitive science is often accused for several kinds of dualism: That between the mind and the body, between external and internal and between individual and social. The separation between mind and body is an inevitable consequence of the decision to simulate the functions of human processes free from their neurophysiological implementation. The separation between internal and external goes hand in hand with the separation of mind and body. These two have led to the assumptions within cognitive science that cognition consists only of mental processes and that these can be studied outside context as general and universal cognitive abilities. Both these assumptions are wrong and the separations are artificial. Their success within the cognitive paradigm probably depends much on that they afford measuring with todays tools and that they fit with the computer metaphor of mind. Hutchins (1995) use the thought experiment the Chinese room set up by Searle to illustrate the point that the present definition of cognition does not capture all there is to it. The story of the Chinese room goes like this: Imagine Searle sitting in a room with a slot in the door. Chinese people come and slip sentences written in Chinese through the slot and Searle slips back other strings of characters which the Chinese take to be clever responses to their questions. To his aid Searle has a basket of Chinese characters and a rulebook that explains which sequences of characters he should concatenate in response to certain strings of characters. Searle does not understand a word of Chinese but he still accomplishes the task with the help of the rulebook and the stored characters. Now, the room behaves as if it understands Chinese yet neither Searle or anything else in the room can be said to have this understanding.
Designing Activity
15
With this reasoning Searle originally intended to prove his claim that syntax is not sufficient to produce semantics in the ongoing debate of whether computers can be attributed cognitive properties. Although the thought experiment is powerful in its original intent as well, Hutchins interpret it in a different way. He describes the Chinese room as a sociocultural cognitive system and the really nice thing about it is that it clearly shows that the cognitive properties of the person in the room is not the same as the cognitive properties of the room as a whole. The cognitive science approach has modelled the cognitive properties of a sociocultural system, which, in this case, would be Searle inside the room. Real cognition is what we can accomplish with our mediational means and any assessment of cognition must take into consideration the mediational means that are used since their mediation is a crucial part of the action under study: Drawing the inside/outside boundary to firmly thus create the impression that individual minds operate in isolation and encourages us to mistake the properties of complex sociocultural systems for the properties of the individual:
If one believes that technology is the consequence of cognitive capabilities, and if one further believes that the only place to look for the sources of cognitive capabilities is inside individual minds, then observed differences in level of technology between a technologically advanced and a technologically primitive culture will inevitably be seen as evidence of advanced and primitive minds. Hutchins, 1995, p.355
The dualism between internal and external is also due to the computer simulation approach. As described above it seemed at first simplest to abstract away from human interpretation of the outer world, in order to simulate problem solving. From a sociocultural perspective, this is not simulation of cognition. Cognition must be something more than mental. It must be seen as cultural and social. The mentalistic focus on mental phenomena is irrelevant because cognition is always dependent upon what sort of cognitive support, social or physical the actor can make use of. One last illustration of this is made by Perkins (1993) distinction between a person-plus and person-solo perspective in education. The person-solo perspective is most common in contemporary education. Taking a person-solo perspective implies that when measuring performance or knowledge of students it is only the mental capacity of the student that is measured, as in an exam with a sheet of questions, a pen and an answering sheet, where all mediational means are removed. Taking a person-plus stance requires one to see the individual only as one part in a functional system or as an agent acting with mediational means, which mirros the every day situation. It is the student with all the mediating structures, such as textbooks, calculators, teachers, friends, family, that are necessary for him/her to perform in daily activities that consists the unit of analysis. The student learns by making use of all these mediating structures. This is also the performing postgraduate situation, except that teachers now are substituted by others such as supervisors, managers and co-employees. The point Perkins makes is that the exam situation in many contemporary schools and universities does not measure the full performance of the individual (person-plus), neither does it measure the skills needed in future working life. It only measures some dimensions of mental capacity, i.e. person-solo. Hutchins (1995) argues that moving the unit of analysis out beyond the skin reveals other sources of cognitive accomplishment. These other sources are however not mysterious, they simply arise from explicable effects that are not entirely internal to the individual. As mentioned the cognitive paradigm is also often accused for a dualism between the individual and the social. This is, for example, apparent in the separation
Theoretical background
16
between individuals and organisations, institutions and culture. We speak, for example, of phenomenon such as the military culture, the atmosphere of a pub or the soul of McDonalds as if these existed apart from the individuals that create them. As described above culture is a process. We actively create and affect our culture at the same time as we learn from and adapt to it. We may speak about culture as being there but it is important to keep in mind that its existence is completely dependent on the individuals that create it and the artefacts that capture it. The atmosphere of my favourite student pub down the block is, for example, very special. This atmosphere would not be the same if the management changed the direction towards something else, if the students that work there did not do it for the pleasure of the job but for the money, or if the people who go there would not be students. This, and many other factors, constitutes the culture, or atmosphere, of this particular pub and there would be no such atmosphere if these people, norms and values were replaced by others. Finally, I will criticise the cognitive approach for taking an individualistic stance. Much research has narrowly focused on the individual in isolation and problems and solutions have been considered individual. Often, however, problems are more general. If one individual have a problem it is likely that others within his/her community also experience this. In the sociocultural approach this is a natural inference since artefacts and social constructs mediate the individual activity. Since cognition is distributed the same must be true for the cognitive problems. Many problems and solutions are in fact bound to the sociocultural context rather than to the individual and this will not be apparent until one shift focus from individuals to activities. The individual never acts alone. His or her actions and thoughts are constantly shaped by the cultural-historical context set up by the activity, of which the individual is part. If I would say, for example, that I can choose what I like to wear when I go out this is, in one sense, a lie. My choice is most certainly affected by present norms and rules, by what I want to achieve, by my relations to those who I will go out with etc. This might seem obvious, and on an abstract level it is. But it must also be obvious, then, that by abstracting cognition and studying individuals outside their everyday activities and, hence, their sociocultural context, we can not understand cognition, just because cognition is situated and bound to the activity.
The cognitive approach can be criticised for relying on rationalism. While the critique of the former three -isms is concerned with the focus of the cognitive paradigm this last one is more concerned with the approach. The fundamental rationalist stance is that the proper solution can be thought out. How can this be rationally solved? This is a typical engineering approach where one tries to define the most important properties of the problem and then devise a solution. The solutions that are sought for are the most rational. The rationalistic approach is topdown in that it starts of with a problem (which has been identified by asking people rather than observing) which is broken apart in to finer detail until a rational solution is found. In order to characterise rationalism it might be necessary to polarise it with an alternative non-rationalistic bottom-up perspective, empiricism. In this case the answer would not be found in ones own rational problem solving, but in the study of the problem at hand. Why is this a problem and how do the individuals involved perceive it, would be natural questions to ask. A consequence of this approach is that the solutions that emerge as an understanding of the nature of the problem, as it is experienced by those who have the problem, might result in other solutions
Designing Activity
17
than the most rational and effective one, but which better fits with current practice and culture. The rationalist approach is hence to think out a solution. This started with Descartes Cogito ergo sum, I think therefore I am. Descartes could prove his being by the fact that he could think. Thinking is the holy process; our whole being is in our thoughts. What Descartes never questioned is why it is that he could think. While the cognitive paradigm would argue that language did not develop until we could think the sociocultural stance is the opposite. From a sociocultural perspective language is the master tool that mediates thinking and it was not until language was invented that humans started to reflect on their own being. The most important part of human development started with the onset of culture, which in turn was a result of the ability to have higher mental process by the invention of tools, which mediate action. Activities can not be rationally thought out since an activity is meaningless without its context. Activities can only be fully understood within their cultural-historical context since they depend on the culture in which they exist and their present form is a result of an evolutionary process that has been fuelled by more than individual intentions. The discussion above has mostly been concerned with the approaches taken in cognitive science. But the view of cognition that underlies cognitive science, influenced by dualism, mentalism and individualism, also underlies HCI and other directions within the cognitive approach.
Theoretical background
18
Designing Activity
19
Chapter 3
THEORETICAL RESEARCH FRAMEWORK
With this chapter I present the theoretical research framework for this thesis. While I want to keep it short and clear I still must deal with the underlying theories in detail as they are the foundations on which this thesis is built, and I will make use of the concepts in the remaining chapters. I will also get back to some important points stated in this chapter later on and relate them to my research questions presented in the first chapter on the one hand, and to the results of the study that will be presented in chapter six on the other. The theoretical research framework employed here views cognition from another perspective than the traditional perspective in cognitive science. I will use the perspective presented here when I collect and interpret my data. Hence, if you want to understand this thesis you must, metaphorically speaking, wear the same glasses as I have had during the course of this study. In my choice of theoretical framework and approach to the study I hope to avoid the traps of dualism, mentalism, individualism and rationalism. I will try to avoid dualism by viewing cognition as distributed over peoples individual minds and the mediating means that these people employ. By taking activity as my unit of analysis I will also try to avoid separating individual action from organisational context and culture. Furthermore I will take measurements to avoid mentalism with the awareness of the distributed nature of cognition and with the help of a conceptual model that brings with it a wider unit of analysis. I will also be careful not to give in for individualism by using the activity as a minimal meaningful context to understand individual actions. Lastly, I will try to circumvent the traps of rationalism by trying to understand practice by focusing on action and separating between reported and observed activity.
20
The contextual approaches all adhere to the sociocultural approach as an alternative stance towards cognition. Although they have their own research agendas they have very much in common. I will, however, not deal with the approaches in detail but only present their major differences. One of the first one to seriously challenge the traditional cognitive science and human-interaction community was Lucy Suchman (1987) with her revolutionary book Plans and Situated Action. This was the beginning of an approach that would be named situated action. In situated action the focus of study is situated activity or practice, as opposed to the study of the formal or cognitive properties of artefacts, structured social relations or enduring cultural knowledge or values. The approach recognises that activity grows directly out of the particularities of a given situation and hence the unit of analysis is the activity of persons-acting in setting, or the relation between the person and the institutional environment (Nardi, 1996c). Cognition is seen as bound to the situation and the activity and cannot be abstracted away or studied outside the context set up by the activity. The other major direction within the sociocultural approach is distributed cognition. The key person behind this approach is Ed Hutchins, who developed the approach together with his research team in San Diego. Key elements are already described in the sociocultural approach to cognitive science presented above, which is based on Hutchins work. Distributed cognition asserts as a unit of analysis a cognitive system composed of individuals and the artefacts or mediational means used. By viewing cognition as distributed, rather than individual, the distributed cognition approach aims to show how intelligent processes in human activity transcend the boundaries of the individual actor.
Activity theory
Activity theory has got its roots in the classical German philosophy and writings by Marx and Engels, but foremost the Soviet cultural-historical theory of activity
Designing Activity
21
(CHAT) initiated by a group of revolutionary psychologists in the 1920s and 1930s. One of those psychologists was Lev Vygotsky. He formulated the basic concepts of the approach together with his colleauges A. R. Luria and A. N. Leontev. Vygotsky introduced the concept of mediation and the X terms artefact and tool. Artefacts are all things that are created. They can be both material, such as a hammer, and abstract, such as purposeful social relations and language. Tools are artefacts that are used to interact with S O or control ones surroundings and mental processes. Since all tools are artefacts tools do not have to be material either. On the contrary Vygotsky considered Figure 2: The subject relating to the object of his action either language to be the most prominent tool ever developed direct with lower mental and the best instrument we have to interact with, process or indirect via the use of structure and manipulate our mind and thoughts. As the a mediating artefact. term action is used here thinking is also considered an action and artefacts are used in all actions. Vygotsky argued that we always relate to something via a mediating artefact (for example, a tool or a sign) when using higher mental processes. The lower mental processes, such as reflexes, relate directly to the object of our actions (Figure 2). For a more in-depth analysis of the development of CHAT and how it has influenced activity theory see Bruner (1985) and Veer & Valsiner (1991). Activity theory is a set of basic principles that constitute a general conceptual system, rather than a highly predictive theory. It can serve as a foundation for more specific theories. The basic principles of activity theory include the hierarchical structure of activity, object-orientedness, internalisation / externalisation, mediation and development (Kaptelinin & Nardi, 1997). I will deal with each of these principles here, accept for the principle of object-orientedness, which will be discussed in the following section when presenting the activity system.
In his doctoral thesis Engestrm (Engestrm, 1987) further developed the activity theory formulated by Vygotsky and Leontev and introduced the graphical model of an activity system (Figure 3). The model contains three mutual relationships between subject, object and community. An activity is actually a systemic whole in the sense that all elements have a relationship to the other elements. The top part form the original model developed by Vygotsky. It describes how tools mediate the subjects (an individual or sub-group whose agency is chosen as the point of view in the analysis) actions toward a particular object. The object often leads to confusions in activity theoretical reports. Object orientedness is however the most important principle (Kaptelinin & Nardi, 1997) and understanding the activity system model requires first of all an understanding of the object.
22
Tools
Rules
Community
Division of labour
Figure 3: The activity system. Based on the conceptualisation by Engestrm (1987). Adopted from Kuutti (1996).
First of all, the term object has two meanings in activity theory and I have tried to avoid one of them in order to avoid the common confusion. The double meanings derive from the translation of the texts and concepts in activity theory from Russian to English. On the one hand the object is connected to the motive in the activity (see Figure 4):
Every activity is directed towards something that objectively exists in the world, that is, an object. For example, a computer program is an object of a programmers activity. Kaptelinin et al., 1999, p. 28. Italics in original.
It is the object that motivates and gives energy to the activity. While actions and operations change in an activity the object remains the same (Kaptelinin & Nardi, 1997). This sense of the object is simply termed motivation in Figure 4 to avoid confusion. The other sense of the object is the difficult one. In this case the object refers to the one in Figure 3. In this case the object refers to the raw material, or problem space, that is moulded and transformed into outcomes with the help of physical and symbolic, external and internal mediating tools (Engestrm, 1999b). Instruments mediate ones actions towards the object. When planning a project, for example, the material that is transformed into an outcome could be knowledge about available resources and constraints. The outcome is a physical project plan on a piece of paper. Spreadsheets, whiteboards, heuristics etc. mediate ones action towards the object, gradually creating a project plan. Engestrm & Engestrm (1990) give a more straightforward example: The blacksmith (subject) uses a hammer (instrument) to mould a piece of iron (object) in order to produce a sword (outcome/motive). Getting back to Figure 3 the subject is also a member of a community. A community is a group of individuals that share the same overall object. Hence a community may consist of anything from a small team to a whole nation. The subjects relation to the community is mediated by the rules that exist within that community. Rules cover explicit and implicit norms, conventions and social relations within a community. Rules constrain actions and interactions within the activity system. The division of labour, finally, refers to the explicit and implicit organisation of a community as related to the transformation process of the object into an outcome. It refers both to the horizontal division of tasks between the members of the community and to the vertical division of power and status. The
Designing Activity
23
division of labour mediates the relation between members of the community and the object (Kuutti, 1996). An example adopted from Kuutti (1996) can help to concretise the model. Imagine a software team programming a system for a client. The object is the not-yet-ready system, which should be transformed into a delivered, bug-free application (the outcome). The team is the community sharing the same object, perhaps joined by some representatives of the customer. There is a certain division of labour: between manager and subordinates, between software developers and user representatives, and between the team members. There is a set of rules covering what it means to be a member of this community. Some of the rules may be explicit and set by law, but many of them are most certainly implicit, either as a part of general working culture or developed as the team works together. In each step of the transformation process different set of tools and instruments are used in the transformation process, for example, analysis methods, computers, programming tools, walk-throughs etc. The collection of tools has got a history. It is a result of a process of accumulation and rejection at both company and team level, and additions and deletions to it may occur during any project. Whatever the members of the team do during the project it is shaped by the context of the activity. At the same time there may be another activity where the object is the financial status of the software company and where the community consists of team managers and their superiors. Every team manager has tools and tricks to keep the project within budget and profitable. There is also a certain division of labour and a certain set of rules, probably different from that within the team. A third activity can be imagined whereby some of the team managers compete against each other for an available position as a department manager. Here the object is the relative weight of capabilities and assets, real or imagined, of each applicant in the eyes of the selection committee. These assets can be transformed by increasing ones own assets or diminishing those of others. Each participating team manager has the relevant tools excellent financial results from the project, for example. Again, there is a division of labour, at least between applicants, those who do the selection and those who can affect the selection. There is also a set of implicit and explicit rules of what constitutes correct behaviour in the situation and what does not. According to Kuutti (1996) real life situations always involve an intertwined and connected web of activities as in the example above and these can be distinguished according to their objects. Participation in connected activities that have very different objects can cause tensions and distortions (e.g. the position of the team manager in the example: bug-free delivery versus excellent financial results) (Kuutti, 1996). The same software team activity will however look quite different if we take the point of view of another subject in the community, for instance a designer. Yet both subjects share the overall object the project and the problems connected to it. An activity system is always heterogeneous and multi-voiced. Different subjects, due to their different histories and positions in the division of labour, construct the object and the other components of the activity in different, partially overlapping ways (Engestrm, 1999b). The division of labour mediates the different members relation to the object and the object has different personal sense for each member, depending of his/her role in the community (i.e. the division of labour).
24
Activities are structured hierarchically. One activity includes a chain of separate actions that are conscious goal directed processes that must be undertaken to fulfil the object (Kaptelinin & Nardi, 1997). The overall activity is directed towards an object-related motive of which the individual subject is often unaware. Each separate individual action is directed towards a particular conscious goal. Actions are in turn realised through operations that are dependent on the conditions in which the activity is performed. Leontev, who continued the work of Vygotsky on the activity theory, points out that activities can never be fully understood by studying individual actions. The individual actions do not make sense if they are studied in isolation and not being seen as being part of a larger activity with a particular motive. If you study a child play a game such as hide and seek you will not understand what is going on if you study the child alone, as he sneaks behind bushes and climbs trees. You must study the whole activity of which the childs actions is a part
Level Activity
(planning a project)
Action
(arranging a meeting)
Goal
(Finding mutually suitable meeting times)
Individual or group
(Team leader)
Operation
(Dialling the phone)
Conditions
(The line is busy)
The arrows between the levels in Figure 4 illustrate that the structure of an activity is not static but rather under constant development and change. Operations are, for example, former actions that have been automated to the extent that they can be performed without conscious control. The classic example within cognitive psychology is the non-conscious actions (i.e. operations) of gear shifting while driving a car. The shifting of gears is performed automatically, without conscious awareness. If the conditions change, however, as when you buy a new car with different layout of the gears, the gear shifting operation can turn into a conscious action with a particular goal. For example, if the gear will not fall into reverse the driver will shift focus from the action of driving, to the action of getting the gear in the reverse. The same action would otherwise have been an operation had there been no breakdowns in the performance, which gives rise to a shift of focus. It is, however, important to notice that the actions are still performed within the activity of driving a car (somewhere). The boundary between actions and activities is also relative. An activity can lose its motive and become an action, and an action can become an operation when the goal changes. What is an activity for one person can be an action for someone else. Kuutti (1996) also points out that one and the same activity can be realised using different actions, depending on the situation. On the other hand, one and the same action can belong to different activities, in which case the different motives of the different activities will cause the action to have different personal sense for the subject in the context of each activity.
Designing Activity
25
Vygotsky coined the term Zone of Proximal Development (hereafter ZPD) in his work on developmental psychology. The ZPD was a definition of the difference in capabilities of the individual in isolation and with the help of, what he named more competent peers. With more competent peers Vygotsky intended parents and other elderly with a similar function. When speaking of development Vygotsky wanted to see to a childs potential achievements with the help of others and by imitating and learning from the more competent peers. Referring to the discussion above of the role of artefacts in sociocultural systems the original formulation of the ZPD can be criticised for not including mediational means as components of the ZPD, as in Perkins (1993) difference between personsolo and person-plus (see page 15). This was recognised by Brown (1993) who redefined the ZPD as:
A zone of proximal development is the region of activity that learners can navigate with aid from supporting context, including but not limited to people Brown, 1993, p. 191
According to the theory put forward by Vygotsky the child internalises the potential knowledge within the ZPD. Internalisation constitutes the transformation of external activities into internal ones by moving through the zone of proximal development. Externalisation constitutes the opposite process of transforming internal activities into external. This is often necessary when an internalised action needs to be repaired or scaled, e.g. when a calculation is not coming out right (Kaptelinin et al., 1999). The important difference between the sociocultural and the cognitive view on learning is that the cognitive paradigm does not point out the importance of how the material and social surrounding mediates development. Activity theory differentiates between internal and external activities. Internal activities correspond to the concept of mental processes in cognitive science. External activities are performed physically. Internal activities cannot be understood if they are analysed separately, in isolation from external activities, because it is the constant transformation between external and internal that is the very basis of human cognition and activity (Kaptelinin et al., 1999). When performing an internal activity one does not use any external mediating means, as when calculating 5x5 without help. Knowledge can reside in artefacts as accumulated cultural experience. When internalising an action the knowledge that previously resided in the tool, and emerged in the use of that tool, is transformed and represented in the person. Wertsch (1998) points out that the term internalisation can be misleading since it implies an opposition between external and internal processes that leads to the kind of mind-body dualism that has plagued philosophy and psychology for centuries (Wertsch, 1998, p. 26). It is also problematic because different parties use it for quite different phenomena. Wertsch further believes that the metaphor of internalisation is too strong since many processes dealt with in cognitive science were never intended to be internalised, as follows in the discussion of the missteps in cognitive science on page 12. Instead he proposes the term mastery. I interpret Wertsch as being against the use of internalisation as a transaction of things (e.g. knowledge) across a border (e.g. the mental - the outside world) rather than a transformation of something external into something else internal. With the latter interpretation of internalisation as transformation follows that the formerly external knowledge or action is not the same as the now internal knowledge or action, but they are most certainly related. Although I agree with Wertsch that another term is preferred I will still use the term internalisation in this report since it is an
26
established term within the activity theory. The reader should however be aware of my interpretation of the term. The concept of internalisation can be illustrated by an example. Imagine a child learning how to ride a bicycle. In order to be able to get rolling he takes help by a pair of supporting wheels. The skill of how to balance the bike is built in to the structure of the supporting wheels. The child uses the tool over and over again until that time that he can ride the bike without the supporting wheels. The skill that formerly emerged in using the wheels as mediational means is now internalised by the child. It is not easy to avoid thinking of the inside-outside boundary as a border with transactions of information passing through. In fact, Reddy (1993) shows that this is not only embedded in our culture but also in our language. There is not room to give an account of Reddys argumentation here but I will only summon the conclusions. Reddy names the predominant metaphor for communication in the western culture for the conduit metaphor. He claims that it is embedded in our language and frames our thoughts (which are mediated by language as an instrument). This is unfortunate since the metaphor gives a wrongful picture of how human communication works that has devastating cultural and social consequences. According to the conduit metaphor language transfers thoughts by allowing the sender to pack the thoughts into words, in speaking or writing, and send them as through a conduit for the receiver to unpack them, in listening or reading, in the other end. The success or failure of communication depends on the success of the speakers insertion process or the receivers unpacking process. However, since unpacking something is a rather passive process, more often the blame is directed to the failure of the sender to insert the information correctly. The conduit metaphor fosters a view on communication as a success without effort phenomenon. In fact, Reddy argues, the opposite is true. Human communication will almost always go astray unless real energy is expended. No one actually receives anyone elses thoughts directly in their minds. In terms of the conduit metaphor, what requires explanation is failure to communicate. In the alternative view on communication that Reddy describes failure is not something odd:
Partial miscommunication, or divergence of readings from a single text, are not abbreviations. They are tendencies inherent in the system, which can only be counteracted by continuous effort and by large amounts of verbal interaction. Reddy, 1993, p.175
The embedded conduit metaphor turns our language into an enemy when trying to avoid the dualistic view, which has dominated since Descartes. As Reddy shows it is very difficult not to use words which imply this separation of internal and external and that things are transacted across this boundary. Recall that tools enable as well as constrain and frame action according to the possibilities given by the tool. Since language is the most powerful and commonly used mediational mean, the embedded conduit metaphor hinders reflections (mediated by language) on other perspectives of human functioning, such as the sociocultural theories.
One of the points made by Wertsch (1998) is that Mediated action is situated on one or more developmental paths. When discussing development Wertsch points
Designing Activity
27
out that it should be recognised that it is not the agent or the tools that develops in isolation but the mediated action itself. The irreducible dialectic tension between the agent and the mediational means implies that even though changing the mediational means (e.g. introducing new tools) may have a greater effect on production than enhancing the individuals skills (e.g. educating), improvements are not attributable to the individual or the tools solely, but to the mediated action as a whole. The same position is held in activity theory as an understanding of the present use of artefacts requires an understanding of the historical development of the activity and the artefacts involved. Artefacts are not just used arbitrarily. They are the result of a cultural and historical development. As Wertsch (1998) points out the mediational means are often designed and developed for another purpose than for that which they are used. If the purposes conflict with each other this can be one reason to breakdowns in activities. The Dvorak keyboard may serve as an illustrating example. The QWERTY (QWERTY for the top five letters on the keyboard) keyboard that is most commonly used today was actually designed to impede performance. When the keyboard was designed in 1872 one of the requirements was that it should slow down typing so that the typewriter keys would not jam. The skilled typists at the time were faster than the keys in the typewriters and hence the keys jammed. In 1936 August Dvorak devised a new keyboard that would improve efficiency. It was truly more efficient and it is still used to set speed records in typing. It is, however, not used, despite all research that indicate that it, besides being more effective, is easy and fast to learn and that it is possible to switch between the QWERTY and the Dvorak keyboard without difficulties. This reveals something about the power that historical context may have in shaping mediational means (Wertsch 1998). When trying to interpret what a community of people are doing, and why they use a specific set of tools, it is important to have in mind that these tools have developed in the specific community over a long time:
technological change is as much a social as a technical process. Instead of speculating on why this technology has not yet been overtaken by something better it seems more productive to look closely at how this existing technology is actually being used. Hutchins, 1990, p. 192
Mistakes are often done by assuming that the design of the current activity and the mediational means employed is old and (hence) inefficient. New technology is introduced to increase efficiency but instead it demolishes social practices by removing the necessary mediational means that have evolved and been refined over a long period of time. The end result may, instead of a better and more efficient functional system, be a non-working system. In some cases some seemingly inefficient way of solving problems may only be the visible part of a complex system of co-ordination among individuals, where the seemingly inefficient instrument that will be replaced by a modern one is the critical mediating tool between these individuals.
Mediation
The principle of mediation is already introduced above and I will only emphasise its importance in activity theory as it is presented in Kaptelinin et al. (1999). First of all, tools shape the way human beings interact with reality as activities are mediated by tools, internal or external. Furthermore, internal activities are shaped by shaping external ones. Secondly, tools usually reflect the experience of other people who tried to solve similar problems before, and invented or shaped the tool
28
in order to make it more efficient. This experience is accumulated in 1) the structural properties of tools and 2) the knowledge of how the tool should be used. The first point is discussed in many theories such as Normans (1998a) discussion of affordances. The second point is, however, more critical for activity theory, which emphasises that a tool comes fully into being when it is used, and that knowing how to use it is a crucial part of the tool.
So the use of tools is an evolutionary accumulation and transmission of social knowledge, which influences the nature of not only external behaviour but also mental functioning of individuals. Kaptelinin et al. 1999, p. 32.
The artefact itself can also represent information explicitly as in a text and implicitly in their structure and shape, as in the example with the calculation above. This information emerges as knowledge in the use of the artefact in the activity.
Contradictions
Activity theory uses the term contradiction to indicate a conflict between two elements in an activity. These contradictions manifest themselves as problems, raptures, breakdowns or clashes (Kuutti, 1996). Contradictions can emerge within activities or as a result of external influence from other activities. Kuutti (1996) describes contradictions as sources of development and a platform from which new learning may evolve. This process can be hastened if the contradictions are articulated and the subjects in the activity get an opportunity to change the current state of affairs. When examining an activity on the level of actions the source of development is breakdowns. Breakdowns hinder a smooth mediation of the tool at hand. For example, in system design an application is said to exhibit a high level of invisibility if the application itself is not apparent but only mediates the action. A breakdown in the action causes the focus to shift from the object that lies beyond the tool to the tool itself. The elimination of breakdowns improves mediation.
In activity theory the unit of analysis is the activity, consisting of a subject (an individual or group), an object or motive, artefacts and sociocultural rules. As pointed out by Kaptelinin et al. (1999), just as we cannot pull apart hydrogen and oxygen if we want to understand water, we cannot pull these pieces apart without violating the very essence of human activity. Understanding human activity requires a commitment to a complex unit of analysis (Kaptelinin et al., 1999). The complexity derives from the basic ideas that the human mind emerges, exists and can only be understood within the context of human interaction with the world and this interaction, i.e. activity, is socially and culturally determined. The activity itself provides the minimal meaningful context for understanding individual actions. The conceptual framework in activity theory and the activity as the unit of analysis forces the user of the framework to analyse action, i.e. doing, in context. Issues that must be dealt with are: What motivates the actor and what is he/she trying to achieve. What role do instruments (not only physical and/or external but also abstract and/or internal) play as mediators in the actors efforts to reach his/her goals?
Designing Activity
29