Some simple definitions in Distributed Cognition (DCog)

Mark Perry

Please do not use without referencing this page. These definitions are drawn from my chapter 'Distributed Cognition' in the book 'HCI Models & Theories', Ed. Jack Carroll, 2003

Cognitive artefact ­ a man-made, or man modified tool to support mental activity. Examples include number systems, slide rules, navigational charts and even language itself. Whilst generally applied to a single individual, within the framework of DCog, a cognitive artefact is also a tool that supports the co-ordination of information processing between entities within a functional system.

Cognitive ethnography ­ a qualitative method of data collection used by DCog researchers based around observation. It is 'cognitive' because it focuses on computational information transformations within a functional system.

Cognitive science ­ the scientific project dedicated to understanding the processes and representations underlying intelligent action.

Computational metaphor (of cognition) ­ Hutchins (1995a:117) defines a 'computation' as referring "to the propagation of representational state across representational media". The computational metaphor is the position taken that cognition a form of computation, and that mental state is encoded in the same way as computer representations: hence the term 'metaphor', as they are deemed to operate in the same way, if not in the same physical architecture. Within cognitive science, the computational metaphor is also known as the 'representational theory of mind', in which computations are actions on representations. DCog claims that the computational metaphor can be applied to a unit of analysis broader than an individual's mind (i.e. the functional system).

DCog ­ the form of distributed cognition developed by Hutchins in the early 1990's. It is distinguished from other uses of the term 'distributed cognition' by its explicitly computational perspective on goal-based activity systems.

Functional system - the functional system is the unit of intelligent activity that is analysed in DCog; it may be composed of a collection of any number of individuals and artefacts.

Information processing ­ within cognitive science, problem solving is seen as an information processing activity (see Computational metaphor (of cognition)), in which encoded information is acted on and transformed in the resolution of a goal held by a cognitive entity.

Propagation of representations ­ this is a component of information processing activity in which sequences of transformations from one representational form (or medium) to another result in information processing activity. Hutchins claims that "Representational states are propagated from one medium to another by bringing the states of the media in to coordination with one another (1995a:117).

Representation ­ A representation can be seen as encoded information, either as a symbolic abstraction of a thing (classical cognitive science), or as a distributed set of nodes (PDP) that, together, have meaning. According to the representational theory of mind, human brains operate on symbolic representations, or codes. DCog extends this to show how transformations to representations need not be entirely symbolic, but may be enacted through manipulations on physical media that have a representational status (e.g. a navigational chart, or a drawing).

Representational state ­ defined by Hutchins (1995a:117) as "a configuration of the elements in a medium that can be interpreted as a representation". Problem solving occurs by successive re-representations of the problem (i.e. a representation of the problem) through a series of intermediate representational states into a solution (i.e. a representation of the solution). DCog researchers attempt to make the representational state of a functional system explicit and document how changes to its representational state result in goal-directed problem-solving activity.

Socially distributed cognition ­ this is the theoretical position that the goal-directed group activity can be understood in computational terms.