Report on the INTERCHI’93 workshop "Rethinking theoretical frameworks for HCI"

Report on the INTERCHI’93 workshop "Rethinking theoretical frameworks for HCI", Amsterdam, 24-25th April, 1993.

Yvonne Rogers, Liam Bannon and Graham Button

In ACM SIGCHI Bulletin, vol. 26, no. 1, Jan 1994, pp. 28-30.

Introduction

This one-and-a-half day workshop was intended to bring together researchers concerned about the state of theory in HCI, to discuss the adequacy of current theoretical frameworks and to examine more closely a number of alternative or extended frameworks that have been proposed for HCI. A further aim was to examine the recent ‘turn to the social’ and its implications for design practice in HCI. A wide variety of position papers (to put it mildly) were received, from which 15 were selected for presentation and discussion at the workshop. These ranged from critiques of the role of theory in HCI, expositions of various theoretical frameworks and the importance of considering methodology in relation to theory. Several authors also described how their alternative frameworks had enabled them to 'open their eyes' to alternative design solutions when analysing particular problems in a work context.

The workshop was organized around three inter-related themes with the intention of engaging in both reflective and projected thinking. These were: i) what is the problem in HCI? (ii) What does my theoretical approach have to offer HCI, and (iii) How does my theory relate to practice? The participants were asked to address these questions in relation to their position papers. A general concern that became central to all themes was what theory was being used for in HCI. Several attempts at identifying and demystifying its role were suggested and it became clear that in fact it was being used in a multitude of ways. These included a background from which to: frame the problem, pose questions, to analyse, to describe and to explain. There was a general consensus, however, that the most ‘scientific’ use of a theory– to propose and evaluate predictions about human performance–was not appropriate for the current wave of alternative theory building. The lessons learnt from attempts to apply information-processing models to user performance were taken as sufficient evidence that the field of HCI is far too rich and complex to force into a set of hypotheses that can be quantitatively tested. Furthermore, the field is too diverse and changing to be formulated as a coherent theory of HCI. Alternatively, the role of theory in HCI should be to inform and guide system analysis and design.

What is the problem in HCI?

While the field of human-computer interaction (HCI) has expanded rapidly over the past 10 years, and our ways of understanding, describing and (occasionally) predicting in HCI have grown, there has also developed a palpable sense of unease - at least within significant sections of the HCI community - about the state of the field. Much of this concern stems from the adequacy of the most commonly used conceptual frameworks in the field to provide a comprehensive and comprehensible account of exactly what constitutes the field, what is the object of study, what are the core problems that either have been solved or remain to be solved, how can we improve the quality of current HCI designs, etc.

The discussion of the first theme "What is the problem in HCI?" was prompted by Liam Bannon noting a few pertinent issues for HCI currently:

1) There has been increasing concern with the adequacy of the standard information-processing model of human cognition, imported from cognitive psychology, as a suitable base for a theory of HCI.

2) One aspect of this concern is expressed in the need to somehow handle "context" - a much-abused term in HCI recently.

3) There is increased awareness of the variability of human action and thus the need for supporting the human actor in the situation, sometimes viewed as "user empowerment" (another much-abused concept).

4)There has been renewed interest in the role played by artifacts in human activity.

5) There has been a growth of interest in accounts that deal with the social as opposed to the individual in relation to work tasks, technology and practices (witness the interest in Computer Supported Cooperative Work, or CSCW).

6) There has been general debate on the role of "theory" in the human sciences.

7) The relevance of any theory for design practice has been seriously questioned.

This was followed by a series of brief presentations of 5 position papers by Kari Kuutti, Steve Draper, Tom Dayton, Markku Nurminen and Dan Shapiro, and followed by a lively discussion among all the participants for the remainder of the first half-day of the Workshop.

In his contribution to the Workshop, Kari Kuutti elaborated on some of the problems of current theoretical frameworks for HCI and argued that an alternative 3-level conceptual framework of activities, actions and operations - what has come to be called "activity theory" (AT), drawn from the work of the Russian psychologists L.S. Vygotsky and A. Leontiev - might provide a more suitable conceptual base for understanding many issues in HCI. As he himself admitted, however, the predictive and practical utility of the theory "remains to be seen", though an example of how the theoretical framework can be used in practical research was presented later in the Workshop by another participant, Osmo Saarelma. Steve Draper provided some useful critique and possible re-working of traditional cognitive science notions in the light of his understanding of AT work in his input papers, and in his oral presentation elaborated on further ways in which our conception of "goals" can be interpreted as evaluation procedures rather than declarative states. In his short position paper, Tom Dayton discussed the general issue of predictive vs. explanatory theories, noting the different criteria that we use for evaluating theories, arguing for the utility of theory even if predictive power is low. This particular point lead to quite an amount of discussion later in the Workshop, as to what constitutes a "theory" and the general consensus appeared to be that the explanatory function of the different theories under discussion was the most crucial aspect for those present. The sociologist Dan Shapiro discussed ways in which the concept of interdisciplinary theory might make sense. While not wishing to "police" theoretical debate, he raised some questions about the relations between theoretical frameworks, and how we might conceive of "interaction" in the HCI term. Finally, the information systems researcher Markku Nurminen discussed the different perspectives that one can adopt in understanding HCI issues, arguing for the "humanistic" perspective which has been the guiding frame for his own research over a number of years.

These alternative conceptual frameworks are not simply abstractions but have consequences for how one frames what the problem is in HCI, and how one would actually go about designing information systems. His contribution lead to further discussion concerning paradigms, perspectives and conceptual frameworks. At the end of the session, the general impression was of having understood some of the conceptual frameworks more fully, and underscoring the role of theory as an explanatory device, and as an aid to understanding, rather than as a predictive instrument. The idea of participants agreeing to a single conceptual framework as to the way forward was never on the agenda, as participants were eager to understand and discuss different conceptualizations and how they might illuminate our understanding of HCI issues.

What does my theoretical approach have to offer HCI?

Judging by the diversity of presentations given in relation to the second theme, it appears that theory building in HCI is becoming very much a growing cottage industry. Alternative ideas, concepts, terms, and methodologies from a range of disciplines are starting to enter the HCI stage. Theories and approaches ranging from distributed cognition (deriving from cognitive science), layered protocol theory (developing from ideas in communication and perception), activity theory (originating from ideas in soviet psychology) to situated action (ideas from sociology and anthropology) are being propounded as having an important role to play.

The emergence of a polyphony of theories in HCI appears to be developing into a healthy form of pluralism. Moreover, a general observation during the workshop was that the 'new' theories, far from being incommensurable–as proponents of ‘pure’ theories often are (e.g the social versus the individual)– have much in common. In particular, their aims and methodologies adopted were similar in many respects. Where they differ, is in the lenses they offer to view the phenomenon, the questions this leads to and the subsequent analyses applied. Each of the researchers, with their varying backgrounds, have derived ideas from both familiar and unfamiliar domains, through which they have developed their own HCI-oriented framework. In doing so they have discovered certain connections and relevances to system design. Be it the conception of user-interactions in terms of actions and operations sliding up and down a scale of consciousness (activity theory), or in terms of propagations of representational states (distributed cognition), the alternative conceptions of HCI has facilitated some illuminating understandings.

Following from the first session, a pertinent question that was asked was the extent to which the nascent theoretical frameworks could provide a better or more extensive account of the HCI phenomenon than existing theory. A suggestion was that they could be more useful in informing design practice. Indeed, the evidence presented, indicated that they have enabled the researchers to discover the tasks and activities that are actually accomplished in the work place in a manner that has important implications for the redesign of the system under observation. Hence, in this context, the theories are primarily acting as heuristic tools - be it activity theory, distributed cognition or other - for systematically and constructively thinking about how to support and enhance working practices.

How does my theory relate to practice?

In the final session, the question as to the relationship between theory and practice was further discussed in terms of how theory related to design practice. Likewise, the question was predominately addressed in terms that sought to reveal the way in which design practice could be informed or transformed in terms of the theory that was being propounded. Suggestions varied but two that were underscored in the discussion were, first, an attempt to make HCI a designed-based science through the development of scenarios, and second, an argument about the need to reduce enforced sequentiality by, for example, designing control mechanisms that allow user initiated operations to be processed immediately.

However, whilst it was the relationship between theory and design practice that was initially developed in the discussion that followed the paper presentations, the workshop turned to addressing the relationship between theory and user practices. One of the papers had suggested that the object of design is not so much computer applications but the work that applications are developed to address, and another had suggested that activity theory provides an understanding of the use made by people of external symbolic memory devices. The question of whether or not the theory appropriately reproduced the observable activities of persons at work was asked and it was suggested that what was at issue was not only a theoretical problem but also methodological problem concerning the investigation of human action and interaction.

Couched in these later terms the perplexing issue of how to build in knowledge of human action and interaction at work into HCI, becomes even more taxing because the status of that knowledge can no longer be taken for granted. Thus it was argued that the turn to the social in HCI is not just a simple matter of introducing socially based concepts into a discipline that is largely dominated by cognitive concepts and then using these concepts to inform and improve design. In the turn to the social a lid has be taken off a boiling cauldron of competing social science perspective, theories, concepts and methods and the task that confronts HCI is not just one of building in the social into its conceptual framework but also one of developing appropriate descriptions of human action and interaction which whilst they need to be based upon the empirical inspection of human doings must also be pitched at an appropriate epistemological level.

Implications

With distributed computing, multi-media systems, and CSCW applications already a reality and the possibility of the development of ubiquitous computing technology on the horizon, phenomena that have been traditionally the domain of social science such as "interaction", "the organisation of work and the workplace", "the nature of organisations", "context", and the like will become increasingly relevant for designers. At the present time HCI, which is still wedded to its cognitivist concepts, is ill-equipped to address these issues. There is a growing recognition that HCI needs to overhaul its conceptual apparatus, so that it can remain relevant to computer design in the light of the new moves that are taking place, a recognition that it was felt had been reinforced by this workshop.

Participants:

Arne Raeithel, Institute of Psychology, Hamburg University, Germany.

Bonnie Nardi, Hewlett-Packard Labs, Palo Alto, CA, USA

Christine Halverson, Dept of Cognitive Science, University of California, San Diego, USA.

Dan Shapiro, Dept of Sociology, Lancaster University, UK

David Frohlich, Hewlett Packard Labs, Bristol, UK.

David Jennings, Consultant, Sheffield, UK

Graham Button, Rank Xerox Cambridge Europarc, Cambridge, UK

John Carroll, IBM Watson Research Center, Yorktown Heights, New York, USA.

Kari Kuutti, Dept of Information Processing Science, University of Oulo, Finland

Liam Bannon, Dept. of Computer Science & Information Systems, University of Limerick, Limerick, Ireland.

Markku Nurminen, Dept of Computer Science, University of Turku, Finland