Distributed cognition; or, cognition within a complex system

…Complex system, because, after all, humans and human organization within a particular environment, such as a functional organization, are complex.

On p. 177 of “Distributed Cognition: Toward a New Foundation for Human-Computer Interaction Research,” the authors claim that the patterns of information trajectories must be stable in order for the underlying cognitive architecture to emerge. I would disagree with this view unless the following view is also consented for discussion. In his book Chaos: Making a New Science, Gleick argues that communication is chaotic, or orderly disordered, and can thus be studied with chaos-theoretic methods. Moreover, Parunak et al. (p. 2) argue that distributed multi-agent systems, such as foraging ants coupling with the environment, should be studied in terms of two systemic levels: micro level, wherein chaos occurs, and macro level, wherein order (social organization) emerges. And order emerges because of the chaotic dissipation of information by the shared environment to which the actors are casually or intentionally coupled. Thus, in my view, there must exist some sort of cognitive architecture even in chaotic communication systems. Either way, both views perpetuate the theoretical claims of distributed cognition, including embodied cognition.

A very important observation of this paper is the strong possibility of creating virtual social/collaborative environments whose properties could replicate and even transcend the properties of real environments (pp. 184-185). If the theoretical framework behind distributed cognition enables such an observation, then workspace awareness in Gutwin’s framework may be subsumed by Hutchins’ framework along with the issues it attempts to address and tackle. The loop Hutchins proposes (p. 181), wherein cognitive ethnography plays a central role, may be carefully employed to address the issue of workspace awareness. The next question we face is “How?“.

While the latter question is yet-to-be-addressed and explored as the authors humbly conclude, it is a legitimate relief to finally observe the harnessing of three disciplines–cognitive sciences, cognitive anthropology, and socio-psychological sciences–to provide the theoretical foundation for distributed cognition so as to incrementally build better interfaces.

It’s all about the framework

It’s probably not about the groupware itself. It’s not about what groupware should be able to accomplish in order to raise workspace awareness. It’s most certainly about the theoretical and methodological framework that churns out the awareness architecture from which instances of effectual groupware are yielded. Gutwin and Greenberg’s descriptive framework of workspace awareness provides a succinct working definition of awareness in collaborative environments. While most detailed depiction may appeal to us as redundant because of common sense, it is important to note that the fine skin of details provides for better insight into efficient interfaces. For instance, the HCI components discussed in the second part of the paper, in which the authors attempt to adapt their framework to distributed teams, may sound to us quite redundant in terms of factual representation, but at their time these components were viewed as essential and radical. It reminds me of Fitts’ Law, thanks to which we have sufficient insight to build better, say, menus.

With respect to the empirical results, I should want to assume a relatively skeptical position. First, it would have been more prudent on the authors’ side to reveal the number of users involved in the experiments as well as the geographical area they belonged to or the cultural-cognitive processes they adhered to, in the sense of Hutchins’ distributed cognition approach and in the sense of ethnographic analyses. It would be interesting to read whether awareness results would change from one organizational culture to another (e.g. North American vs. Russian). Moreover, research shows that when analyzing work settings, it is crucial to familiarize oneself with the work practices involved in such settings, which wasn’t the case with the authors in their experimental design, or that’s what the reader is left to conclude.

It is, however, important to note that workspace awareness in Gutwin and Greenberg’s sense seems to provide certain benefits in dynamical environments (systems). One would be team proactivity, wherein individuals are able to cope with unintended issues for the overall benefit. Another would be resilience of the dynamical environment (distributed team) itself. These two properties, however, are not bounded by groupware per se, but are properties that may very well be studied by humanizing the CSCW methodologies. In light of that, it would also be interesting to study how such frameworks would apply to geographically distributed teams as well — this would be more realistic in today’s working environments.

I should want to press on one final point, which seems to be capriciously particular of me. This entire idea of workspace awareness framework seems to me to be based on the more generic idea of distributed multi-agent systems (cf. p. 5, Neisser’s model). Here, the environment is considered “alive” and dissipates information, whereas the agents are supposed to be proactive and resilient while digesting such information. I argue that CSCW research needs to go after such guidelines while avoiding focus on narrowly constrained domains of study.