Last week I decided to give (=was coerced into giving) a talk at Reboot10. I was apprehensive about talking because I didn’t know what to expect from this audience and felt I wouldn’t have anything interesting to say that they’d like to hear about! Well, I pondered this for a little while and realized that I could simply introduce them to the theory of distributed cognition (dispelling misconceptions and misunderstandings about it) and try to relate this framework to how I hope to study the “social web.”
The accompanying slides can be found here:
The video of the talk will be up shortly (I’ll update here).
What is Cognitive Science?
I started by discussing the field of Cognitive Science (my department at UCSD). Cognitive Science is broad, interdisciplinary field that is rooted in three main areas: brain, behavior, and computation. For example, it studies how people think, what goes on in their heads, how they behave in the world, and (on the computational side) how to model some of these real human traits to develop new theories or hypotheses. Of course, Cognitive Scientists do a number of other things, too. (Human-computer interaction is also informed by studies of Cognitive Science).
What is Distributed Cognition?
Ed Hutchins and other researchers developed the theory of “distributed cognition” (dcog) at UC San Diego in the late 1980s. Dcog is really an approach or theoretical framework for studying all of cognition—it is not a type of cognition. It is also not the same as “collective intelligence” or “wisdom of the crowds,” and it is not simply studying how people interact with each other. It recognizes that “human cognition is always situated in a complex sociocultural world and cannot be unaffected by it” (Hutchins, 1995, p.xiii), and therefore considers people’s interactions with others, with material artifacts in the environment, and with the cultural influences of a place to understand how a cognitive system operates. This cognitive system, or “unit” of analysis, may be at the level of a single individual or it may include many operators and environmental factors (think inside an airplane cockpit or on the bridge of a navy ship).
Distributed Cognition on Navy Ships
Ed’s seminal text on Cognition in the Wild (1995) seeks to understand how people manage to sail large navy ships. The “unit” of analysis here is the entire collection of “objects” involved in navigating a ship: including all the relevant actors (main navigator plus the people who assist him), their tools (gyroscopes, maps, compasses), and the navy culture that imposes various rules and regulations in the way people can (and must) interaction with each other, for example. Studying a setting like this makes it fairly obvious that people are, in fact, embedded in complex environments—saying that “cognition” is what goes on inside the skin or skull of a single individual here overlooks too many other factors that (directly and indirectly) influence what gets done.
Onto the Social Web
This framework can also be used for studying the Internet. The Internet is clearly distributed (from the servers that support the principle architecture to the people who interact on various web services). The “social web” is a term some people are using to describe the increasing social nature of people’s online experiences. Many websites are being built on an inherent social framework, where people can easily connect, communicate, or share with others, and often as a by-product of their primary use of the site. Applying dcog to the social web here, we must study the entirety of people’s online experiences: from their social interactions (exchanges with friends, buddy lists they may have), to the websites they use (interface elements, “social objects” they create and discover from blog posts to social bookmarks), and to the web culture itself (people’s inherited or adopted social roles, rules for sharing or interacting, etc.)
A number of researchers have studied social networks and how people interact on individual websites (Digg, Flickr, Twitter, del.icio.us, etc.), but I believe this is the wrong “unit of analysis.” Today when people use the web, they are actively involved on multiple services even for simple tasks like reading the news. Sites like FriendFeed, SocialThing, and lifestream.fm, which aggregate users’ diverse online activities, suggest that this is increasingly the case. My FriendFeed account, for example, provides a single view (or “stream”) for my actions online: uploading photos to Flickr, posting status updates to Twitter, checking in at a new location on Brightkite, Digging an article, or writing a blog post. Yesterday alone it had about 10 updates of my various activities. Of course, people can choose whether or not to use these types of services to record their histories, but considering I had 10 “items” of activity yesterday, imagine what else I did that wasn’t recorded using this service?
Think about what your morning routine is like. Do you visit certain services to read the news or check updates from your friends or family someplace? The hypothetical (yet realistic) example I gave in the talk was: imagine you start by checking your Gmail account, then flip over to your news reader (e.g., NewsFire). From there you find an article from the New York Times which you read and decide to bookmark on Ma.gnolia (or del.icio.us )—either to share with friends or with plans to revisit in the future. You may decide to check updates on Twitter, from which you discover a friend has posted pictures from a recent trip. You follow the link to Flickr and browse the photos. While this episode may occur over 15-20 minutes, 5-6 websites were used to stay up-to-date on current events (whether personal or public). Of course, your routine may include different websites, communities, or interests—this example is intended to be representative only in its broad structure. However, distributed cognition can be used to study the interactions of single individuals as they operate across a range of online environments, sites and services, cultures and contexts, specifying the correct boundary of analysis to be web-wide, not simply system-wide (i.e. single services).
We could also apply dcog theory to study the flow of information across the social web. Considering just a portion of the hypothetical morning routine from above, a single article was published on the NYTimes.com, appeared in a user’s feed reader, and ended up as a bookmark (pointer, reference) on Ma.gnolia. Perhaps new users later discovered this information through the bookmark. How does this general “topic” (content of the article) relate to similar news stories being published online? In other words, what is an information topic’s history, flow, and influence? Social web services are facilitating the propagation and (re-)production of information (whether original items or “memes”). Here distributed cognition could help direct our study of the context and consequence of information as it exchanges hands and flows across multiple services on the web. (Do social web services help extend the “reach” of information topics?)
Conclusion
In summary to what I thought would be a much briefer recap of my talk, I intend to focus my research on users’ extended (distributed and embedded) interactions with people, interfaces, and networks across the web. Ultimately this will probably involve both log file analysis on a large scale and ethnography of fewer numbers of individuals. It’s interesting to study the social web both from the perspective of the individual users and of the information itself (social objects getting created, shared/propagated, and discovered). Both people and objects make up the web, and distributed cognition theory will help establish the appropriate level of analysis for the specific question we’re answering, by including important factors like social interactions across different web services to subtle cultural influences.
