The Art and Science of Everything

Journal paper summarizes 12 years of ethnographic and experimental work developing distributed cognition as a theoretical foundation for understanding how collectives of people and artifacts accomplish tasks that the authors define as cognitive.

The “central hypothesis is that the cognitive and computational properties of systems can be accounted for in terms of the organization and propagation of constraints.” This reminds me too much of planning algorithms in artificial intelligence and the assumptions of a preconceived, axiomatized system that is being operated in. 

A distro cog system can be a group of people working with artifacts, or it can be a single person working with artifacts, or even what goes on within the single cognitive unit. The last is not a contribution of dcog as much as it is dcog incorporating work like Minsky’s and human information processing theory that divides the mind up subprocessing units coordinated to perform cognition.

Artifacts, including representations, are seen as things that  can be used symbolically or reappropriated according to their physical properties to in performing cognition. Here, authors point out that this means that the real-world baseline for face to face communication is shortsighted, undercutting potential of media spaces. The human body is in the realm of physical things that can be used, and felt through, in imagining, thinking, and remembering (authors talk about navigators feeling directions in relation to bodily orientation). Other people can also be used in coordination to perform cognition as well.

Dcog sees culture as a bunch of partial solutions to problems that people frequently face in the world. What I like about this is that it is an account of why people won’t just adopt the most efficient tool for the job, or even necessarily search for the optimal tool for a job. Culture (their experience in the world) offers them a set of good enough solutions to appropriate to their ends. This speaks to discussions Dan, Robin, and I had about Notebook and Spreadsheets and how people used pretty simple, common schemas for figuring things out. 

This perspective can be described as “emergentist…on many key phenomena” (178)  because the body, world, and brain adapt to one another. In this sense, it is definitely embodied cognition.
Key principles of dcog are:

• people establish and coordinate different types of structure in their environment
• it takes effort to maintain coordination
• people offload cognitive effort to the environment whenever practical
• social organizations can improve dynamics of cognitive load balancing

Dcog is also about a methodology of naturalistic investigations of cognition, particularly using participant observation and video to capture people interacting with each other and the environment, often coupled by experiments to pin down the cognitive activities identified “in the wild.”

The questions suggested by the research framework include how to make representations more active so they help users and supporting people’s conceptualizations of what is going on and what “ought” to be done — both are about cognitive efficacy and instrumental good. There is a third question that I’m into which is “how do we design representations to facilitate their flexible use?”, though I’d broaden that to how do we design technological artifacts to support flexible, ensemble, interactive use — assemblages!

Questions:

• There’s a tendency to describe ethnography as “wild” and “natural” that makes me wonder whether dcog has done reflexive investigation of how its methods shape its observations and what its limiations are.
• Dcog seems to shine when you’re talking about coordinated tasks. What happens when not everyone in a dcog system knows they’re being enlisted in a task? (In a sense, you could say artifacts are like this, and I suspect there’s a use to thinking of people this way too.)
• What happens when tasks aren’t so cut and dry, like people are researching stuff on the internet that they’re interested in and might need later, but they might not, or people are trying to have a pleasant social interaction?
• How much of a claim of universality does dcog make? The language seems fairly careful aboutnot claiming universality without raising the question: “a set of core principles that widely apply” (181)
• How does dcog talk about learning? They say history of use informs our interactions with artifacts (187). That’s something, but not as much as other theories suggest. (But this is less universalist that idea of fundamental cognitive/perceptual affordance, as (I think) Norman suggests.) In dcog, the activities they discuss are all about load balancing, supporting memory, and making satisfactory calculations. Little is said about how and why learning occurs (here, activity theory offers the account of externalization and internalization; communities of practice offers identity and legitimate peripheral participation).
• Dcog also treats information as a thing - “cognitive processes involve trajectories of information (transmission and transformation) so the patterns of these info trajectories, if stable, reflect some underlying cognitive architecture” (177). Is this the same as information as data? Then what are the limitations on the sort of emergence that dcog can talk about usefully (race, gender, culturally slippery concepts, miscommunication and breakdown)? It seems like information is at least a thing in the moment when it is reified as such, though the system has to adapt when breakdown occurs. How does dcog talk about errors and misunderstanding?

It seems like dcog is most effective when you think of things as tasks. So I guess it isn’t surprising that Evans and Chi “information assimilation” studies of social bookmarking, newsreading, and information behavior didn’t find dcog particularly useful in illuminating behavior.

This entry was posted on Tuesday, September 30th, 2008 at 2:38 pm and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.