A framework for the study of external representations in collaborative design settings

Abstract

The use of modeling, simulation and visualization techniques in scientific and technical domains has led to the co-existence of a large diversity of external representations that, when deployed in collaborative work settings, can be designated by the term “shared representations”. This contribution will focus on dyadic cognitive and triadic semiotic perspectives on the issue of interpretation and construction of shared representations. We propose a typology of external representations as a basis for further study of shared representations. As illustrated by examples from an existing corpus, the framework allows describing co-occurrence and superimpositions of different types, as well as transitions from one type to another. In the final section, we anticipate the exploitation of the framework in the light of the design of collaborative systems.

Introduction

The use of modeling, simulation and visualization techniques in scientific and technical domains has led to a large diversity of graphical modes of expression, also called visualizations, such as charts, diagrams, schemas, digital images and so on. For example, in engineering practice, 2D wireframes, matching the contemporary practice of technical drawings at the time of appearance, evolved to become today's 3D realistic renderings progressively including engineering knowledge, trade rules and inferences. Such representations play an increasingly important role in work and training settings in which they support individual task performance and mediate collaborative problem solving across disciplines within design teams.

Research in cognition has shown that different isomorphic representations of a common problem or underlying abstract structure can cause radically different cognitive problem solving behavior, referred to as the representational effect. In a theoretical framework on distributed representations, Zhang and Norman focus on the way in which a problem structure and rules are spread across representations which are either internal, i.e., in the head, or external in the environment (Zhang and Norman, 1994, Zhang, 1997). Their research on puzzle-like problems, such as tic-tac-toe or the Tower of Hanoi, showed how several representations of a unique abstract problem structure actually differ in the degree to which they embody the rules of these games and thus cause differences in problem solving performance. Besides the distribution of a problem structure amongst internal and external representations, the representational effect has been studied focusing on the verbal–visual distinction, such as in Larkin and Simon's (1987) analysis of sentential and diagrammatic representations and Stenning and Oberlander (1995) comparison of linguistic and graphical representations. However, little is known about the effects of using essentially different graphical representations and the dimensions that might underlie performance differences.

The representational effect naturally suggests that there are important consequences related to the choice of a particular representation for a particular task. In collaborative design, selecting a representation has often been a question of mere convenience for the individual engineer in a particular task situation. However, new forms of organization, such as concurrent engineering, coin issues related to the coexistence of several dedicated representations and thus the choice of an appropriate one in co-operational phases. For example, is multi-trade or multi-disciplinary co-operation better supported by a minimalistic representation based on an alleged common background across domains or by a rich representation which accumulates multiple domain-specific aspects? The first option may reduce the embedded information to such a low level that most of the message disappears. The second option may lead to confusion due to the mix of multiple semiotic references. An alternative approach would involve assisting collaborators in defining their own representational formats and having them disclose a variety of perspectives on co-created representations. The design of new collaborative interfaces could well benefit from knowledge of both individual and collective interpretation and construction of a variety of (combinations of) old and new modes of expression provided that essential categories are distinguished within the whole diversity or representations. Thus, in addition to the internal–external and verbal–visual distinctions, a third contrast, namely individual in opposition to collective representations, has to be examined under the assumption that there are fundamental differences between representations in these different categories.

In this article, the focus is on the concept of shared representations in the collaborative design activity and on the suppositions that underpin the notion itself of “sharing”. In fact, we call into question the prevalent intuition that there is a relative easiness with which representations are shared (see also Scaife and Rogers, 1996). The counter intuition would be the suspicion that the proliferation of representational formats might just as well lead to a “confusion of tongues” such as during the construction of the Tower of Babel. A fundamental issue is whether we can distinguish different categories of representations, different “types of tongues”, with different consequences for constructing collective understandings. We start by a review of existing notions that have been imported or developed in the field of design studies, such as the notions of internal and external representations, distributed and coordinating representations, intermediary and boundary objects. What is lacking is an analysis of representations per se, i.e., in their quality of being a representation. In effect, the multiple uses of the term representation beg the question of ontologies of representations founded on the above mentioned distinctions. We build a theoretical frame founded on both external and design cognition, and on cognitive and semiotic perspectives on representation. We then build a typology of representational types and illustrate it with some examples from an existing corpus. In the final section, we present some implications for the design of cooperative systems for collaborative design.