IT 695 - Week 5 - No Brain Is An Island

Distributed cognition. I sort of understand the concept, but I found the articles this week very difficult to plow through. I felt like Uma Thurman in some movie (I can't remember which one), when she said, "I'm reading 2 books. This one, and a dictionary so I can understand this one." But instead of a dictionary, I found myself googling again and again; I swear, even Keen would have been using Wikipedia to try to make sense of some of these articles! Basically, what I get is that cognitive processes are not completely internal. The process involves not just what's going through our heads, but also our environment, the social context and artifacts, our perception, our history.

In his article Distributed cognition: A methodological note, David Kirsch identifed six assumptions that guided his research:

1. We act locally and are closely coupled to our local environments.
2. We externalize thought and intention to harness external sources of cognitive power.
3. Economic metrics have a place in evaluating distributed systems, but they must be complemented with studies of computational complexity, descriptive complexity and new metrics yet to be defined.
4. The best metrics apply at many levels of analysis.
5. Coordination is the glue of distributed cognition and it occurs at all levels of analysis.
6. History matters.

He goes on to explain these assumptions in depth. Closely coupled entities are defined such that a change in one leads to a change in the other, and so on back and forth. Actions may be pragmatic or epistemic, but often are some combination of the two. Kirsch's example described people playing Tetris. They must think about how the piece will fit into the available space, but often this thinking aided by the pragmatic action of rotating the piece repeatedly.

Economic metrics are explained using the example of 2 coffee houses performing the same services but employing 2 different methods - one writes the customer's order on the paper cup, one uses a computerized ordering system. In this example, Kirsch describes how the routines of each coffee house work in terms of speed accuracy, error type vs. frequency, error recovery rate, variance, learnability, and drink complexity. There are literally countless variables that affect whether a routine works or doesn't work, and only by observing the actions and tying them to cognition and computational analyses can a system be measured or explained.

Coordination seems to be the key, since any distributed system is only as good as the combined function of its parts. Kirsch states, "In distributed systems the success of the whole depends equally on all these acts of local choice adding up, working together to move the system closer toward system goals." He strongly advocates the use of simulation and modeling to improve systems, with some cautions about relying too heavily on them. History is invaluable as well, because it is virtually impossible for the elements at work in a system to be without history that has helped changed and define them.