Sunday, October 02, 2005

Rogers and Scaife on DC

There's a good (1997) article on Distributed Cognition by Yvonne Rogers and Mike Scaife that touches on some general assumptions. I've quoted liberally here because I think it's that good. (These two also have an article on Activity Theory (1997) that I think I should read, and one called "A conceptual framework for mixed reality environments: designing novel learning activities for young children" that sounds promising for our Augmented Reality work.
General properties of cognitive systems

A general assumption of the distributed cognition approach is that cognitive systems consisting of more than one individual have cognitive properties that differ from those individuals that participate in those systems.

Another property is that the knowledge possessed by members of the cognitive system is both highly variable and redundant. Individuals working together on a collaborative task are likely to possess different kinds of knowledge and so will engage in interactions that will allow them to pool the various resources to accomplish their tasks. In addition much knowledge is shared by the individuals, which enables them to adopt various communicative practices (e.g. not having to spell out every time they meet someone what they know about a practice, procedure or state of affairs).

Another important property is the distribution of access to information in the cognitive system. Sharing access and knowledge enables the coordination of expectations to emerge which in turn form the basis of coordinated action.
and summaries of a few of DC's classic articles
Case Studies

i) Hutchins (1995) Cognition in the Wild

An example provided by Hutchins (1995) of a distributed cognition analysis of a cognitive system is the navigation of a ship. Here, his focus is on the cultural-cognitive processes that take place when steering a ship into harbour. At a micro-level of analysis, Hutchins describes the detailed coordination of representational states across media that take place for the relatively simple, but critical coordinating activity of plotting a fix. This involves several members of the navigation team taking and plotting bearings of the ship as it comes into the harbour at regular intervals of every 3 minutes or so. It is a highly routinized activity, requiring the complex coordination of people and artefacts - all of which is crucial for ensuring the ship is on course. At a macro level of analysis, Hutchins also describes how these coordinated activities of plotting a fix provide a structured experience for the team members enabling more generally, individual learning of procedures and the cultural practices of the navy. As noted by Hutchins (1995, p374):

"...since most learning in this setting happens in the doing, the changes to internal media that permit them to be coordinated with external media happen in the same processes that bring the media into coordination with one another. The changes to the quartermasters’ skills and the knowledge produced by this process are the mental residua of the process".

Hutchins goes into great detail analysing how the various representational state are propagated across media for this collective navigation activity and in so doing show how ‘the properties of this computational system are as much determined by the nature of the representational media and the pattern of interconnection among representations as they are by the cognitive properties of the individual actors". (Hutchins, 1992, p.2).

This distinction is critical for the distributed cognition approach, emphasising again the importance of focusing on the distribution of cognition through analysing the interactions between the different ‘components’ (i.e. the changes in representational state) of the system over time and place.

2) Hutchins and Klausen (1995) study of cognition in the cockpit

The study analyses the interactions of internal and external representational structure and the distribution of cognitive activity among members of a cockpit flight crew. The analysis shows a pattern of cooperation and coordination of actions among the crew which is viewed at one level as a structure for propagating and processing information and at another level a system of activity in which shared cognition emerges.

3) Rogers (1992, 1993) study of engineering practice

Rogers carried out a study of how networking technology has changed the working practices of an engineering company. Through doing a Distributed Cognition analysis she was able to reveal various breakdowns that occurred in the work activities and the mechanisms by which the group had adapted their working practice to overcome them.

4) Halverson’s (1995) study of Air Traffic Control

Halverson carried out a study of how air traffic controllers interact with a radar system when controlling air traffic. From her observations and analysis she was able to make recommendations of what was important to retain of the existing design with a view towards developing future automated decision-making tools for the controllers.
and a bit on Methodology that suggests what sorts of studies can use DC

The distributed cognition approach uses a number of methods: from detailed analysis of video and audio recordings of real life events, to neural network simulations and laboratory experiments. The type of methodology adopted depends on the unit of analysis that is being adopted and the level at which the cognitive system is being explained.

For cognitive systems that are being described at the ‘work setting’ level it is imperative to carry out extensive field work and become familiar with the work practice. This entails observing the work, making copious field notes, recording events and then transcribing and encoding these. An important part of this kind of ethnographic analysis is re-representing the raw data collected at different levels of abstraction and detail, focusing on the changes in representational state in the cognitive system. Theoretical analyses are also carried out in relation to the assumed properties of distributed cognitive systems.
and some discussion that makes me question (again) whether it's the right framework to look at the role of space in learning. If "place" can be considered an artifact, then I think DC can work. If it can't, I may need to find somehting else.

One of the key questions often asked about the distributed cognition approach is how does it differ from a traditional cognitive science explanation of human activity. Furthermore, what leverage is gained from giving an account of collaborative activities in terms of ‘propagation of representational state across media’? In support of the distributed cognition approach it can be said that it provides a framework and analytic methodology for examining the interactions between people and artefacts which is not possible with traditional approaches to cognitive task analyses. In doing so, it can highlight the complex interdependencies between people and between people and artefacts in their collaborative activities, which in turn, can lead to a better understanding of why, for example, seemingly trivial breakdowns in the communications and interactions between them can have significant and sometimes drastic consequences.

A challenge for the distributed cognition approach is how to integrate concepts from the social and organisational sciences with the cognitive analysis of ‘representational states’. In particular, it is difficult to combine macro-level theories, such as organisational learning, with micro-level detailed descriptions of the intersubjectivity that goes on between two people during a two second encounter. The goal is to find the appropriate level of analysis and explanatory description for the problem that is being addressed.


Post a Comment

Links to this post:

Create a Link

<< Home