Brown on me on d-cog

Wed 07 Oct 2009 12:13 PM

Matthew Brown has a forthcoming paper on science as distributed cognition (d-cog). He gives a generous amount of attention to my discussion of the issue.

In my paper, I characterize d-cog as meaning that a cognitive task is implemented by a process which is not contained inside one thinker's epidermis. As Matt notes, I'm relying here on the task/process distinction and also on an unanalyzed sense of what it is to be a cognitive task. I do this because it lets me give a concise definition of d-cog. The downside of this definition is that giving a d-cog analysis of anything requires specifying its task, and it is not always clear how to specify the task of large-scale scientific practice.

Matt suggests that we might do better to consider a three-fold distinction between operations, actions, and activities. Operations are the automatic, unconscious things we do in performing an action. Actions are short-term, conscious, goal-directed processes. Activities are broader complexes which develop over time.

To apply the distinction to one of my examples: A carpenter might perform the action of cleaning and calibrating a machine. The particular way that the carpenter does this will involve many operations. (The boundary is somewhat fuzzy. If the carpenter does the same action every night for years, then maybe it's just an operation.) The goings-on of the carpentry shop altogether are an activity. The shop's operations might be motivated by the love of carpentry, the desire to make a profit, the desire to do something useful, &c. - and these motivations may shift and change over time.

Matt's point is that the actions and operations have well-defined tasks, and so may be analyzed in terms of the task/process distinction. The activity does not have a well-defined task, and so can't be. My definition of d-cog precludes the activity's being d-cog, and so my definition is inadequate.

He concludes:

So, is science a distributed cognitive system? ... Magnus has challenged it on the basis of whether there is a particular task that science carries out. But what is a cognitive system anyhow, even in the traditional sense of "cognitive system?" This shouldn't stand or fall on the details of a certain framework of cognitive analysis. After all, presumably, I am some kind of cognitive system, even though I am not built to carry out one specific and well-bounded task, even though my cognitive activities ... aren't always as well-bounded as certain cognitive theories might presuppose.
I am not entirely sure what to say, but here is what I'm tempted to say:

Most systems do many different actions. I have no objection to calling the aggregate of actions an activity, but I don't think the broader activity will clearly be cognitive or non-cognitive. The goings-on of the carpentry shop will include some cognitive processes, but they will also include some clearly non-cognitive ones like actually sawing through a board. The goings-on of a human body include cognitive processes like talking and non-cognitive ones like digestion.

I don't have an analysis of what it is to be cognitive. Nevertheless, I think we can recognize some tasks as cognitive (like addition) and others not (like breaking down complex carbohydrates). So a specific action or operation is cognitive if it carries out a cognitive task. A system is cognitive if it involves some cognitive actions or operations.

Matt is a cognitive system because some of his processes execute cognitive tasks (to put the point in my preferred idiom) and because the activity of his life involves actions and operations that are cognitive (to put it in his).This is a pretty thin definition, and it's compatible with my insistence that d-cog can only be made precise in terms of the task/process distinction.

Of course, his paper ends with a suggestion rather than a knockdown argument. He points to rather than elaborates a different sense of d-cog. If it can be worked out and can usefully direct work in science studies, then that would beat the logic-chopping I've done here. We would agree, I think, that the proof will be in the doing.


from: Matt Brown

Wed 07 Oct 2009 01:11 PM

P.D., I think you're pretty much right, here. I've got no ultimate desire to demarcate the cognitive and the non-cognitive, especially at the level of activities and systems (rather than more local cognitive acts and operations). Still, I think it would be profitable to try to give more robust accounts of cognitive activities and cognitive systems in the d-cog framework. (I'll put it on the to-do list!)

More importantly, I think you're spot on: the proof of the pudding is in the eating, and the proof of the utility of d-cog for science studies is in the details of the studies it informs. That's why Nancy Nersessian's new book has recently found its way onto my Amazon Wish-list. ;-)

from: P.D.

Wed 07 Oct 2009 08:55 PM

Matt: Thanks for the comment. I think we just differ on terminology.

If some "d-cog" isn't d-cog in the sense I characterize, then I'm not sure what the label means. And where I can't puzzle out what it means, I don't see the value of applying it. The work - like Nersessian's - can still be interesting and insightful independently of the label.