Schaerding, Austria, 20 - 23 May 1997
INTERACTIVE DESIGN AND ACTIVITY THEORY
Edward Gould
Department of Business Systems --- University of Wollongong, NSW, 2522, Australia
E-Mail: Edward@pcmail-bld40.uow.edu.au
1. Introduction
The main purpose of interface design is to assist users in their professional activity. In order to accomplish this they need to work "through the interface" (Bodker, 1990) to complete the tasks which achieve the goals associated with an activity. Although this is the conceptual province of psychology very little use has been made of psychology in practical interface design (Carroll, 1991). The attempts that have been made at developing user models appear to have suffered from a lack of application to real life problems. This can possibly be attributed to their foundations in the information processing structure of cognitive psychology. This paper elaborates an approach to interactive design based on the Russian developed Activity Theory which provides a more complete analysis of human nature and which avoids the problems inherent in the view of humans as exclusively information processors. Activity Theory is becoming more widely recognised as a promising alternative to HCI problems particularly now that a new book "Context and Consciousness: Activity Theory and Human Computer Interaction" edited by Bonnie Nardi appeared last year (1996). It suggests that the activity theory approach may allow us to move HCI research out of the laboratory and into the field and may also be used as a means of guiding field studies.
The basis for this research arose from a problem encountered in designing an interface for demographers to assist in the onerous task of population modeling. It became obvious that much of the previous design was based on what the designer thought "looked OK" rather than on any sound principles of user interaction with professional activities. What was in fact needed was a better psychological base on which to build a user model.
2. Current Cognitive Based User Models
The diverse nature of the many current techniques used in HCI design and evaluation makes the job of classifying and identifying them as models, methods or architectures difficult. Although most of the original authors referred to their techniques as models, in most cases the word technique or tool is probably a more accurate description. Over the last fifteen years a large number HCI design tools and techniques have been proposed. Howes (1994) has identified twenty three different approaches, each oriented towards a small and manageable part of the overall issue of how people interact with computers.
The best known of HCI tools is GOMS, developed by Card, Moran and Newell (1983) involving Goals, Operators, Methods and Selection rules and is based on the premise that users act rationally to attain their goals. The implications of this are that the user's behaviour can be predicted by determining the `small' number of information processing operations underlying that behaviour. It also assumes that behaviour can be described as a sequence of these operations and that the time a user requires to act is the sum of the time taken for these individual operators. Not surprisingly this method of predicting efficiency is based on the cognitive information processing model of human behaviour approximated by what Card et. al. call the model human processor (MHP). The MHP is a simplification of psychological theories and empirical data intended to be used by non psychologists to predict user behaviour in HCI. More specifically, Johnson (1992) has highlighted a number of limitations with the GOMS model which affect its usefulness. They include its restriction to skilled task performers with no problem solving behaviour and the difficulty in using various aspects of the approach. A more serious criticism by Johnson is the problem of the way the MHP (model human processor) reduces the whole of psychological theory to four stores and three processors. This, along with the way the way in which predictions are derived from an additive model is, according to Johnson, a gross over simplification. Similar criticism has been levelled at an extension to the GOMS technique known as Cognitive Complexity Theory (CCT) which was developed by Keiras and Polson (1985) and a cognitive architecture produced by Young, Green and Simon (1989) called a programmable user model (PUM).
Modelling the cognitive resources of a user has also been adopted as the basis of a framework for HCI analysis by Barnard (1987). It is based on a theory called Interacting Cognitive Subsystems (ICS) where a number of special purpose processing systems are linked together in a cognitive architecture. The human input stimuli of perception followed by cognition then action are assumed by the architecture to be capable of analysis by discrete information processing modules. Johnson (1992) believes that this technique provides a much richer description of human-information processing than that provided by GOMS although it is much more difficult to apply due to its degree of fidelity to human systems. Johnson says: `even for a psychologist to identify consistently and completely all the different subsystems and their correct associations for any given task would be a difficult undertaking'.
In spite of the diversity of models available to designers none have proved to be highly successful for the task. We now consider the common problems associated with these cognitive based models.
3. Problems With Cognitive Based HCI Models
Given that HCI is a study of the relationship between humans and a machine it would seem that a fundamental basis of HCI design should be psychology. Despite this, evidence exists (Card, Moran and Newell, 1983; Polson and Kieras, 1985) that even when sophisticated attempts have been made to extend cognitive psychology in the direction of HCI they have had no effect on design systems. Alternatively, some of the most successful user interface design work of the past twenty-five years has made no explicit use of psychology (Carroll, 1991). Landauer (1991) has also criticised the inadequacy of cognitive science when it comes to providing a blueprint for HCI design. He says that in spite of 100 years of investigation into human behaviour `nothing remotely resembling what one would hope for as a basis for HCI, nothing with substantial generality, power, and detail at the required level of cognition has ever materialised'. In a study of British software houses Bellotti (1988) found that designers were not using any of the methods recommended by HCI research and in fact, most were unaware of their existence. Other criticism can be found in Suchman (1987), Ehn (1988), Bannon (1991), Bodker (1991), Kuutti (1992), Draper (1993), Kaptelinin (1992,1994), Aboulafia, Gould and Spyrou (1995).
Clearly the design of interfaces needs to be conducted in the field, but, this will only succeed if sound psychological structure is applied to the study of people in their work environment. As Carroll (1991) suggests this would allow the incorporation of understanding the user interface in terms of human motivation action and experience. Activity Theory as an alternative approach has been suggested (Kaptelinin, 1992, 1994, Kuutti, 1992) as it can provide the structure for a new approach based on Carroll's suggestion. Activity Theory is based on the cultural-historical traditions of Russian psychology and is outlined in the next section.
4. Activity Theory: The Concepts
The founder of Activity Theory, Alexey N. Leontiev (1978) regarded an individual's activity as dependent upon their place in society, the conditions that make up their lot in life and how this lot is worked out in unique individual circumstances. `Society', he said `produces the activity of the individuals forming it and that activity therefore can only have a structure that is created by given social conditions'. The general theoretical approach he outlined was first to discover the structure of human activity and use it for investigating the essential features of consciousness. In building a user model we need to deal with specific activities, each of which answers a definite need of the subject. According to Activity Theory each need is directed toward an object of that need, is extinguished as a result of its satisfaction, and is produced again in other altogether different conditions. Activity types can differ among themselves depending on the difference between their objects. The object of an activity Leontiev (1978) says is its true motive, which may be either real or imagined, thus activity does not exist without motive. He called a process an action if it is subordinated to the representation of the result that must be attained, that is if it is subordinated to a conscious purpose. He goes on to say that just as the concept of activity is related to the concept of motive then the concept of purpose is related to the concept of action. Human activity he explains, does not exist except in the form of action or a chain of actions with one and the same action accomplishing various activities and may be transferred from one activity to another. This means that activity is usually accomplished by a certain complex of actions subordinated to particular goals that may be isolated from the general goal. In other words, these actions are seen as goal-directed processes but, in society, these processes are rarely related directly to the goal where even the simplest division of labour leads to 'goal-isolation'.
So ther are several different levels of analysis which make up the theory of activity. Associated with each level is a specific type of unit, the first of which is the unit of "activity"(detyatel'nost). Here "activity' means the specific real activities as opposed to human activity in general and includes such things as labour, formal education and play. The next level of analysis is the action (deistvie) which, although related to an activity at a particular time can vary independently of it and is transferable to other activities. The third level of analysis is the operation (operatsiya) and this relates to the concrete conditions under which the action is carried out and includes the resources for reaching the goal. The goal, which is embedded in a specific set of conditions, comprises the task of the activity. This can be summarised schematically (Wertsch, 1985):
5. The Role of Tools
Leontiev (1978) saw tools as both external or physical (artefacts, instruments and machines) and internal or psychological (laws, signs, procedures, methods and language). Technical tools are designed to manipulate physical objects (e.g. hammer) while psychological tools are used by humans to influence other people or themselves (e.g. concepts, advertisements, the calender) (Kaptelinin, Kuutti and Bannon, 1995). Tools therefore are social objects with certain modes of operation developed socially in the course of labour and are only possible because they correspond to the objectives of a practical action. A basic tenet of Vygotsky's theory was that tools occupy a mediating role, they mediate human reaction with the world. Since psychological tools are included in the process of behaviour they alter the entire flow and structure of mental functions. Similarly the physical tool (including computers) alters the process of natural adaptation by determining the form of labour operations (Wertsch, 1994).
6. Activity Theory And Human-Computer Interaction
This is elaborated by Kuutti (1995) who nominates three perspectives where activity theory can play a role in broadening the scope of research and design in HCI. The first is in addressing what he refers to as `the ubiquitous fragmentation of the field'. The second perspective he suggests is that interaction should be studied embedded in social context thereby making use of activity theory constructs such as actors, mediation, historicity, constructivity and dynamics. In the third perspective he recommends that the potential of activity theory's action-operation dynamics be exploited to reveal how the formation of new operations, sense-making and the creation of new activities can be included into the design of interfaces.
Summarising this we can see that Activity Theory provides a paradigm for the description and understanding of the way humans interact with computers within the context of the user's environment. This makes it ideally suited to the study of human-computer interaction. It stresses the vertical integration of psychological analysis and stresses the following points usually missed by the cognitive approach:
Other considerations have been examined, for example Suchman (1993) has shown that interactions with information technology are embedded into logically structured sequences of interactions with other objects and with people. As well as this user behaviour also does not remain static. The normal progression for humans on repetitive tasks is to begin as a novice and progresses through to an expert. The use of an artefact can also change the nature of the task requiring new artefacts to be developed (Carroll et al, 1991). Finally the whole dimension of the meaning of "user" is changing from an individual to a group or indeed the whole organisation (Kaptelinin 1994).
7. Summary
Within this Activity Theory framework we can see that the activity of a person via structure gives us a broader view of a user in the HCI paradigm. In this structure the computer appears as a tool for achieving human goals (motivation). Motives are related to real needs in the user's environment and are located out of the laboratory. In this way the approach allows us to bring together the environment as an integral part of HCI. There are two consequences of this approach. First, in order to create the user model it is necessary to work with the user(s) in their environment in such a way as to discover the motives underlying their professional activity. In other words collaborating with the user in a non-laboratory situation becomes an imperative (it is their needs not the designers that are important). Second, correct tool (interface) design can only be achieved by specifically examining the structure of the activities being performed so that the correct balance of operations between human and machine can be determined. In this way goal formation can be encouraged by successive improvement of the tool. This is aimed at assisting the user achieve the motives previously identified.
Relationship of the Activity Theory Model to Interface Design
In the current demography interface study, real work with demographers was undertaken to discover the motives underlying the professional activity associated with population modeling. An investigation was carried out into the structure of pre-existing activities and the tools associated with earlier attempts at assisting in the process. This included techniques users had developed for modifying populations in specific areas based on a bewildering amount of data output in a very unfriendly tabular form. By examining these user techniques goals could be associated with the motives. Once the goals were established various tools were examined as possible contenders for achieving these goals. It was discovered that many operations performed manually by users could be transferred to the computer and be performed on screen given that suitable interface techniques were made available. A suable prototype has been developed (Gould & Verenikina, 1995) and various modifications have been made to aspects of the tools (sliders, gauges, population boxes, maps, colors, etc.) to assist goal achievement.
References
Aboulafia, A., Gould, E. and Spyrou, T, 1995, Activity Theory vs Cognitive Science in the Study Of Human Computer Interaction, Proceeding of 18th Conference for Information Systems Research in Scandinavia, Goeteborg, Sweden, August, pp 29-38.
Barnard, P., 1987, Cognitive Resources and the Learning of Human-Computer Dialogues, In Carroll, J.M., (Ed), Interfacing Thought: Cognitve Aspects of Human-Computer Interaction, MIT Press, Mass.
Bannon, L.J., 1991, From Human Factors to Human Actors. In Greenbaum and King, (Eds), Design at Work, Erlbaum, NJ.
Bellotti, V., 1988, Implications of Current Design Practice for the Use of HCI Techniques, in Winder, D.M.J. and R., (Eds), People and Computers IV, Cambridge University Press, Cambridge, UK.
Bodker, S., 1991, Through The Interface: A Human Activity Approach To User Interface Design, Lawrence Erlbaum, Hillsdale, NJ
Card, S.K, Moran, T.P. and Newell, A., 1983, The Psychology of Human-Computer Interaction, Lawrence Erlbaum Associated, Hillsdale, NJ.
Carroll, J.M. (Ed), 1991, Designing Interaction, Psychology at the Human-Computer Interface, Cambridge Universtiy Press, Cambridge, UK
Draper, S, 1992, Activity Theory: The New Direction For HCI?, Intl Journal of Man Machine Studies, Vol 37
Ehn, P., 1988, Work Oriented Design of Computer Artifacts, Abets-livscentrum/Almqist & Wiksell Intl, Sweden.
Howes, A., 1994, Learning To Use an Interface: Articulating The Space of Models, Proceeding of CHI'94 Research Symposium, Boston, Mass.
Johnson, P, 1992, Human Computer Interaction, Psychology, Task Analysis and Software Engineering, McGraw Hill, London.
Kaptelinin, V., 1992, Human Computer Interaction In Context: The Activity Theory Perspective, Proceedings of the East-West Intl Conference on Human-Computer Interaction, St Petersburg, Russia, ICSTI, Moscow, pp 7-13.
Kieras, D. and Polson, P.G., 1985, An Approach to the Formal Analysis of User Complexity, Intl J of Man-Machine Studies, Vol 22 No 4, pp 365-394.
Kuutti, K, 1992, HCI Research and Activity Theory Position, Proceedings East-West Intl. Conference on Human-Computer Interaction, ICSTI, St Petersburg, Russia, pp 13-22.
Kuutti, K., 1995, Activity Theory as a Potential Framework for Human-Computer Interaction, In Nardi, B., Context and Consciousness; Activity Theory and Human Computer Interaction, MIT Press, Cambridge, Mass.
Landauer, T.K., 1991, Lets Get Real: A Position Paper on the Role of Cognitive Psychology in the Design of Humanly Useful and Usable Systems, in Carroll, J.M. (Ed), 1991, Designing Interaction, Psychology at the Human-Computer Interface, Cambridge Universtiy Press, Cambridge, UK
Leontiev, A.N., 1978, Activity, Consciousness and Personality, Prentice Hall, Englewood Cliffs, NJ
Suchman, L.A., 1993, Centres of Coordination: A Case and Some Themes, NATO ARW on Discourse, Tools and Reasoning, Nov., Lucca, Italy
Suchman, L.A., 1987, Plans and Situated Actions, The Problem of Human machine Communication, Cambridge Universtiy Press, Cambridge, UK.
Wertsch, J.V., 1994, The Primacy of Mediated Action in Sociocultural Studies, Mind, Culture and Activity: An International Journal, Vol 1 No 4, Fall, Laboratory of Comparative Human Cognition, UCSD, Ca.
Wertsch, J.V., 1985, Vygotsky And The Social Formation of Mind, Harvard University Press, Harvard, Mass.
Young, R.M., Green, T.R.G. and Simon, T., 1989, Programmable User Models for Predictive Evaluation of Interface Design, In Bice K. and Lewis C., Human Factors in Computing Systems, CHI'89, Wings for the Mind, Adison-Wesley, Reading, Mass.
Back to List of Position
Papers