Masterclass 3 Practical Distributed Cognition
Christine Halverson
Social Computing Group, IBM Research

Abstract. Distributed Cognition has been in use in CSCW for over a decade now, yet there remains confusion about what it means and how exactly to do it. Drawing on an excerpt from commercially available video participants will get hands on experience in doing a distributed cognition analysis. In addition we’ll discuss how such an analysis can translate to design.

Introduction. Designing and implementing effective collaborative technologies is complex and inherently interdisciplinary in nature. Many current theories used to understand work practice tend to be singular in their outlook – focusing on either the cognitive, the social, the organizational or other restricted frame of reference. While such approaches are often well developed (cf. CHI literature) they can overlook important issues. In response, CSCW has picked up several theories that provide a more comprehensive approach. Two of these are Activity Theory and Distributed Cognition.
This masterclass focuses on Distributed Cognition, a theoretical framework developed by Ed Hutchins and colleagues at UCSD in the late 80s and 90s (Hutchins, 1995). Its theoretical and methodological underpinnings are a hybrid of cognitive science, cultural and cognitive anthropology, and other social sciences. As a tool, its intended use is to analyze and account for complex, socially distributed cognitive activities.
The focus of the early studies by Hutchins and colleagues have been work settings where a diversity of technological systems and other artifacts are a key part of what gets done. (Such as, (Hutchins and Klausen 1992; Hutchins 1995; Hutchins 1995)) This makes it well suited to being applied to designing, evaluating and implementing CSCW systems.
Theoretically distributed cognition starts from the assumption that cognition (cognitive work) is literally distributed. That is, it doesn’t just reside inside the head. One of the things that make humans special is our ability to construct social, cultural, and technical systems that taken together help us achieve tasks than we could not do unaided. (How often do you do your daily tasks sitting naked in a room using your mind alone?) One advantage of the theory is that it allows for a flexible unit of analysis. This flexibility works well in CSCW contexts as it provides a mechanism for scaling the scope to adjust for complexity, time, and focus.
Collaborative work is both social and cognitive, and both need to inform the design of collaborative computing systems. CSCW by necessity contains artifacts and distributed cognition provides a theoretical language that can explore all of these on an equal footing.
Goals and content: The goal of this class is to introduce the theory of distributed cognition, situate it with respect to other approaches in CSCW, and provide some hands on experience doing an analysis.
The tutorial will demonstrate application of the theory to design and evaluation in CSCW contexts using examples drawn from Hutchins and the instructor’s research. Then, as a group exercise we will go through the experience of working through the principles presented by analyzing a common video source to determine the requirements and design of a collaborative system to support the communication and coordination needs of a distributed team. Video source is an excerpt from a movie so the video and audio are clean and easy to understand. In addition participants can do the exercise without domain specific knowledge. However, pulling supporting information from other sources to supplement direct observations will be discussed. Materials will include a step-by-step walkthrough of a previous analysis, as well as analysis of the video segment.

Intended Outcome. At the end of the day I want the students to be able to walk out feeling that they have the tools and confidence to try a distributed cognition analysis in their area.

Target Audience. Anyone interested in exploring another methodology for analyzing CSCW settings. Some experience with fieldwork (or having previously taken Randall and Rouncefield’s course in Fieldwork and Design) will be useful. Should be useful for social scientists and technology builders from a variety of domains.

About the presenter. Christine Halverson was introduced to distributed cognition by working with Ed Hutchins at the University of California, San Diego from 1990-1995. Her dissertation used distributed cognition to analyze the effects of introducing new technology into a function of air traffic control. She is currently a researcher in the Social Computing group at IBM T. J. Watson having worked at IBM (off and on) for over a decade. 1999-2000 she worked at SRI designing ubiquitous applications using speech for interaction. Prior to 1995 she did joint work with NASA at NASA-Ames Research center and at various field sites. She has used distributed cognition, in some form or another, in a number of domains including Air Traffic Control, securing auto loans, online communities, and most recently studying programmers of highly parallel computers.  
Attendance is limited to 8 or so participants. Expressions of interest should be sent to Christine Halverson (krys@us.ibm.com) to confirm that this class will fit your needs.

Bibliography.
Hutchins E, Klausen T (1992 sub, 1996 pub). Distributed cognition in an airline cockpit. In: Communication and Cognition at Work. Middleton D Engestrom Y. Beverly Hills, CA., Sage Books.
Hutchins E (1995). How a Cockpit Remembers Its Speeds. Journal of the Cognitive Science Society 19.
Hutchins E (1995). Cognition in the wild. MIT Press, Cambridge, MA.