Projected Cognition – extending Distributed Cognition for the study of human interaction with computers William H. Edmondson * , Russell Beale School of Computer Science, University of Birmingham, West Midlands B15 2TT, United Kingdom Received 14 May 2007; received in revised form 8 September 2007; accepted 23 October 2007 Available online 7 November 2007 Abstract In this paper, we introduce the notion of Projected Cognition as an extension to Distributed Cognition. Distributed Cognition is a conceptual framework which can be useful in studying human interactions with artefacts; the idea is that of cognition not bounded by the cranium but instead perfusing artefacts in ways that are recoverable. We argue that this analysis has not been fully understood in relation to the behaviour of humans with artefacts in that the intentionality in behaviour has been ignored. We argue that we need to view the human as sometimes projecting their intention in behaviour onto the artefacts they use, and suggest that this conception permits greater clarity in the study of user behaviour with artefacts such as computers. We illustrate the development with case studies of two users of complex configurations of computers as well as examples drawn from the published literature. We conclude with consideration of some design implications and discussion of related domains in HCI where Projected Cognition could be influential. Ó 2007 Elsevier B.V. All rights reserved. Keywords: Distributed Cognition; Observational study; Multiple computers; Virtual desktops; Projected Cognition; Computer supported cooperative working 1. Executive summary This paper presents a new idea – Projected Cognition – by means of which analysis of human interaction with arte- facts, especially computers, can be made more comprehen- sive. The paper is essentially in three main parts. The first part (Section 3) reviews our understanding of the widely used concept of Distributed Cognition. Distrib- uted Cognition acknowledges that a purely intracranial conception of cognition is too narrow: cognition can be distributed over time, over space and objects, and over other cognizers. It is proposed that work in such a tradition has become insensitive to the intentions of users and that these intentions can be key to understanding what users are doing, especially in complex settings. This proposal is made in the context of discussions of various well-known examples from the literature on Distributed Cognition. Projected Cognition is promoted as a way of shifting ana- lytical focus to the intentions of users. It can be envisaged as an aspect or subset of Distributed Cognition and has the important characteristic that the person distributing their cognition is doing so with intent, or in such a way as to per- mit recovery of intent. Projected Cognition is seen in func- tionality which is allocated to or imposed on resources. The second part (Section 4) looks in detail at the com- puting resources used by two people who have each arranged their work environment around more than one computer. One user has four computers and six screens in his office, and his work covers a variety of different activ- ities: experimental data analysis (number crunching); teaching; general administration; specific administration (admissions work as part of undergraduate recruitment activity). The second user has only two computers, but his utilization is both rigidly patterned and flexible: he con- siders he has two major tasks and each requires (different) configurations of the two machines as means to deploy the resources he needs. Observational work coupled with face- 0953-5438/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.intcom.2007.10.005 * Corresponding author. Tel.: +44 121 414 4763; fax: +44 121 414 4281. E-mail address: w.h.edmondson@bham.ac.uk (W.H. Edmondson). www.elsevier.com/locate/intcom Available online at www.sciencedirect.com Interacting with Computers 20 (2008) 128–140