D-cog in d-machine

Mon 17 Apr 2006 01:46 PM

This is the last entry written in a Hungarian cafe. It's a revised version of the d-cog paper, and this aside that did not make the cut:

Some authors distinguish between collective cognition and distributed cognition, both of which are distinct from individual cognition. In individual cognition, the cognitive activity is done by one person and the representations are all contained in the mind of that person. In collective cognition, several people are involved and the task is not merely the sum of their individual cognitive tasks. In distributed cognition, the task is carried out by one or more people along with cognitive artifacts like chalk-boards, computers, and whatall else.

The distinction does not strike me as being important, and I do not think we need jargon to mark it. As I use the phrase, distributed cognition (d-cog) includes all the non-individual cognition: cases that involve artifacts and also those that involve multiple agents but no artifacts. Two reasons:

First, the insight of d-cog is that a system can perform a task that we think of as a cognitive task without any of the agents involved having the big-picture task in mind. This happens when there are material artifacts like scratch paper and hoeys, but it also happens when people unreflectively and collectively arrive at a solution to a problem. Calling the former distributed but the latter collective overlooks the important similarity.

Second, both kinds of cases involve representations outside of the agents' minds. When I calculate a large product on a chalkboard, the black-and-white representations are enduring. When a couple engages unreflectively in the kind of cueing that allows them to remember as a couple more than either might remember alone, they trade more transient but equally real representations. This phenomenon is known as 'transactive memory', and I think it may count as d-cog. In any performance that is not merely an aggregate of several peoples' individual cognition, some representations will be traded around.

In the paper, I cut through some of jargon wrangling by using `d-cog' rather than `distributed cognition' when I mean the thing that I am talking about in the sense that I have specified. If the locution catches on and comes to mean something more diffuse, perhaps I could distinguish just the thing I have in mind with the phrase distrizzibuted cognishizzle.