Human Computer Interaction
ERCIM News No.46, July 2001 [contents]

Supporting Collective Design Activities: A Cognitive Task Analysis Method

by FranÁoise Darses, FranÁoise DČtienne, Pierre Falzon and Willemien Visser

The EIFFEL research group at INRIA carries out research activities in the domains of Cognitive Ergonomics and Cognitive Psychology. Their aim is to model both individual and collective design activities and to assess and specify tools and methodologies supporting design.

Studies on reasoning in design have usually been carried out on individual problem solving activities. In response to the increasing need to assist collective work in an industrial context, more recent studies have shifted their focus. A major concern in industrial modernisation is the creation of new organisations which support collective work, greater interaction between designers and manufacturers, as well as capitalisation and reuse of design knowledge.

Our working hypotheses are:

Our methodological approach is as following. We conduct empirical studies, either field studies or laboratory experiments. Our empirical studies examine two kinds of design situations:

The former studies concern upstream research on particular activities which need to be supported. They allow us to construct a reference model of the activity. Based on this kind of model, special support needs are identified. The latter studies allow us to assess different kinds of technological systems or methodologies.

COMET: A Cognitive Task Analysis Method for Collective Design
One example of the research contributions of the Eiffel group is a Cognitive Task Analysis (CTA) method for analysing and modelling collective design: COMET. The ensuing objective is to build cognitive models useful for intelligent assistance systems for designers.

The design of products is mainly achieved by teams, in which designers must cooperate to develop the solution. As any other professionals, designers need assistance tools, which can be either computational (CAD or CAM for instance) or methodological (project management tools, decision making methods, etc.). To be efficient, these tools must be able to support the cognitive processes that underlie the designersí activities. Thus, collective design cognitive processes must be identified and modelled. But it appears that more often than not, the cognitive dimensions of designersí work are directly extrapolated from observations of their activity or from interviews, without going through a phase of cognitive modelling. This approximate nature of the analyses makes cognitive system design more like the result of inspired innovation rather than the result of well-grounded methods. CTA methods are thus needed to understand design activities and to develop tools for designers.

COMET is based on the analysis of the dialogues exchanged between designers in co-design situations. COMET gives some methodological guidelines related to a cognitive ergonomics approach of the activity. These guidelines have been elaborated and used in two case studies of co-design situations: computer network co-design and software review meetings. COMET has also been applied and assessed in other design studies.

COMET takes place when the socio-technical macro-analysis of the activity has finished: the ergonomic problem to solve has been formulated, the concerned people have been identified, the prescribed task has been understood through interviews, the organisational demands have been highlighted. COMET stems from observation of work in situ, and especially from observations of design meetings. It provides a frame for analysing design dialogues recorded in such situations. The frame is made up of two distinct and hierarchical levels of coding: basic level (where individual UNITS are identified) and composite level (where the ëCo-operation Movesí are identified).

At the basic level, utterance turns are cut up into one or more individual UNITS according to a coding scheme developed on a Predicate ó Argument basis. Predicates correspond to actions (ACT) implemented by participants; arguments (OBJ) correspond to objects related to the action. Each unit is modulated (MOD), according to the form of the predicate (Assertion or Request). Thus, each UNIT is coded as MOD[ACT/OBJ].

At the composite level, frequent and consistent UNITS are grouped into sequences. Such sequences can be formed in various ways, on a qualitative or on a quantitative basis, ie on a concept- or a data-driven basis. Sequences correspond to basic co-operative interactions called ëCo-operation Movesí. COMET was developed for analysing co-design processes occurring in the same place, where the designers are present in the same room. It can also be applied for design work mediated by communication tools (such as web-based tools), where the designers do not design at the same time. The growing importance of such cooperative design situations makes COMET a good candidate as a CTA method usable for such cooperative design environments.


Please contact:
Francoise.Detienne — INRIA
Tel: +33 1 3963 5522