Thinking some more about Sam Adkins’ description of the distinction between EPSS and workflow learning, I decided to go back and refresh my memory on the concept of distributed cognition. As Gary Dickelman and Jan Greenberg note [PDF],
Distributed cognition looks for cognitive processes, wherever they may happen, on the basis of the functional relationship of elements that participate in this process (Hutchins & Hollan, 1999). A process is not cognitive just because it happens in a brain; conversely, a process isn’t deemed noncognitive because it happens in the interactions among many brains. The second distinguishing principle concerns the range of media assumed to participate in cognitive processes. Hutchins and Hollan note that traditional information-processing theories look for cognitive events in the manipulation of symbols inside an individual’s head. Distributed cognition looks for a broader class of cognitive events; it does not expect these events to be encompassed within an individual’s head. Hutchins and Hollan state that if these principles were applied to the observations of human activity in its natural state (e.g., how people go about their jobs on a daily basis), at least three kinds of cognitive process distributions would become clear: first, cognitive processes may be distributed among members of a social group; second, cognitive processes may involve coordination between internal and external (environmental and/or material) structure; and third, cognitive processes may be distributed through time in a way that the end results (products) of earlier events can change the nature of events that come later. field notes, record work events, and transcribe and encode them. The ultimate goal is to abstract the artifacts and process elements of the native human activity system and replace them with new representations that are computer based. If successful, the distributed cognition system will be re-represented in the performance-based system, such that a human being engaged in the system will be able to achieve goals without support or “training events” that are external to the system. According to Don Norman- The sciences of cognition have tended to examine a disembodied intelligence, a pure intelligence isolated from the world. It is time to question this approach…. Humans operate within the physical world. We use the physical world and one another as sources of information, as reminders, and in general as extensions of our own knowledge and reasoning…. With a disembodied intellect, isolated from the work, intelligent behavior requires a tremendous amount of knowledge, lots of deep planning and decisionmaking, and efficient memory storage and retrieval. When the intellect is tightly coupled to the world, decisionmaking and action can take place.
Two points here. First, Sam’s requirements for bi-directional components and feedback loops might be productively understood and explored as examples of distributed cognition. In other words, if we map how an organization needs to collectively “think” about a business process then we can tease out where we need those bi-directional components and feedback loops. Second, I think that when people invoke “emergent learning,” what they may really mean is distributed cognition. Both concepts share a collectivist picture of thought as a group activity. However, unlike emergence, distributed cognition does not explicitly disallow meta-awareness by the participating individuals. (In fact, Gary and Jan explicitly explore this idea in their article.)
Beyond the Dickelman/Greenberg piece, if you want to read more about distributed cognition then check out Don Norman’s book Things That Make Us Smart. While it’s not nearly as well written as his classic The Design of Everyday Things, it has an extensive and highly readable section on distributed cognition.