Distributed Cognition: Toward a New Foundation for Human-Computer Interaction Research Interaction Research

A cognitive theory seeks to understand the organization of cognitive systems.

Distributed cognition extends beyond the individual to encompass interactions between people and with resources and materials in the environment.

Example: the memory processes in an airplane cockpit involve a rich interaction between internal processes (i.e., in one person’s brain) and external manipulation of objects and representations of traffic. The material world allows the distributed cognitive system to be reorganized to make use of a different set of internal and external processes.

1. Distributed across members of a social group. Issue: how can we more effectively support complex tasks, mediate networked interactions and manage and exploit the ever-increasing availability of digital information?
2. Involving coordination between internal and external (material or environmental) structure.
3. Distributed through time such that earlier events transform the nature of later events.

1. Elaborate a distributed cognition approach
   • Socially distributed cognition
   • Embodied cognition
   • Culture and cognition
   • Use of ethnography
2. Describe how the theory of distributed cognition may provide a new foundation for HCI.
   
3. Describe how the theory can help the design of new digital work materials.

Socially Distributed Cognition

Studies of various group processes (e.g., jury decision making, economics, scientific endeavors) have lead to the conclusion that social organization is itself a form of cognitive architecture.

Distributed cognition means more than that cognitive processes are socially distributed across members of a group. It includes phenomena that emerge in social interactions.

Embodied Cognition

Minds are not passive representational engines, whose primary function is to create internal models of the external world. The relations between internal processes and external ones is more complex, involving coordination between internal processes – memory, attention, executive function – and external resources – objects and artifacts that surround us.

The human body and material world take on central, rather than periphery, roles because the organization of mind is an emergent property of interactions between internal and external resources.

Adaptive success is as much in the complex interactions among body, world and brain as in the inner processes bounded by skin and skull.

Culture and Cognition

The study of cognition is not separable from the study of culture. Culture emerges out of the activity of human agents. Culture in the form of a history of material artifacts and social processes shapes cognitive processes.

Traditional cognitive science views culture as a body of content on which the cognitive processes of individual persons operate. Distributed cognition recognizes that culture shapes the cognitive processes of systems that transcend the boundaries of individuals.

The environment is a reservoir of resources for learning, problem solving and reasoning. It provides us with intellectual tools that enable us to accomplish things that we could not do without them. BUT, culture may blind us to other ways of thinking, leading us to believe that certain things are impossible when they are in fact possible, if viewed differently.

Ethnography of Distributed Systems

Since the cognitive properties of systems that are larger than an individual play out in the activity of the people in them, a cognitive ethnography must be an event-centered ethnography.

We are interested not only in what people know but in how they go about using what they know to do what they do.

The meanings of actions are grounded in the context of activity. This context must be captured for the activity to be studied. Expertise in the domain under study is important.

An Integrated Framework

The authors; proposed approach to HCI research contains the following elements:

• Distributed Cognition
• Cognitive Ethnography
• Experiment
• Work Materials
• Workplaces

Distributed cognition identifies core principles, such as:

• People establish and coordinate different types of structure in their environment
• It takes effort to maintain coordination
• People offload cognitive efforts to the environment whenever practical
• There are improved dynamics of cognitive load-balancing available in social organizations.

The principles identify classes of phenomena that merit observation and documentation. They also point to experiments that can help illuminate the impact of changes of parameters that theory states are important.

Example: the introduction of a new work material is a form of ethnographic experiment, which allows the theory to be tested and revised.

Cognitive Ethnography seeks to determine what things mean to participants and how those meanings are created.

• Often structures are used in ways not anticipated by the designers of tools.
• Experts often make use of environmental structure to simplify tasks. Example: pilots display the test pattern on weather radar as reminder that final fuel transfer is in progress.

• An experiment is a socially organized context for cognitive performance. Most cognitive theories seek to explain experimental data.
• Experiments are important because the complexity of real world settings places limits on the power of observational methods.
• Experiments can refine the theory of distributed cognition that in turn can be applied to improve design.

Work Materials and Workplaces

• Since the design process creates new tools for workplaces, there are new structures and interactions to study.

Navigation aboard US Navy ships

• Outcomes that mattered to the ship were not determined by the cognitive properties of any single navigator but were the product of the interactions of several navigators with each other and with a complex suite of tools.
• Some insights:
  • Showed how a single computational description could cover two systems with radically different representations.
  • Showed that the cognitive processes required to manipulate a tool are not the same as the computations performed by manipulating the tool.
  • Showed how learning occurred both at the individual and organizational level.
  • Example: bearings are understood not just as 3-digit numbers but also that navigators feel a bearing.
  • Observations could be used as the basis for the design of electronic charting tools.
• Distributed cognition has been applied to airplane cockpits and air traffic control. Using theoretical interpretation of ethnographic findings, one author developed a graphical interface to the autoflight functions of Boeing 747-400.

• By paying close attention to how people actually exploit real environments, and describing those phenomena in appropriate theoretical terms, it may be possible to go do better than just imitate face-to-face communication.
• Consider symbols used as tokens to refer to some real world entity. Ethnographies show that people shift back and forth between attending to the properties of the representation and the properties of the thing represented, sometimes intentionally blurring the two. Example: representation of a fuel gauge on a flight panel, flick gauge to get needle to move..
• In GUIs, we are supposed to feel as if we are manipulating the real objects themselves, not just the stand-ins. Example: when we drag a file icon from one folder to another, we think about moving the file, not just the icon.
• Sometimes we’re just working with the representation, without considering the real object. Example: moving icons around on desktop. We’re not moving the underlying objects.

1. How can we design representations to facilitate their flexible use?
2. How can we make representations more active so that they help users see what is most relevant when deciding what to do next?
3. How can we shift the frame of interpretation to a achieve a better conceptualization of what is going on and what ought to be done?

• The history of use of a tool can sometimes inform our interactions with that tool. Example: most recently used pieces of paper occupy the tops of piles on our desk.
• The side effects of use often provide resources for the construction of expert performance. Sometimes these side effects are designed out of “clean” and “simple” digital work environments. Example: cockpit of Airbus A-320 (no details in this article).
• Digital objects could encode information about their history of use. Examples: copy history of source code could help find errors. Including who edited sections of code on scroll bar.

• Goal: move beyond mimicking mechanisms of earlier media.
• Supports exploration of dynamic multiscale interfaces.
• Multiscale spaces are particularly appropriate for hierarchical information.
• Create and interact with structured information based on a zoomable interface.
• Zooming and panning are the fundamental interaction techniques.
• Facilities include portals to support multiple views, lenses to filter and furnish alternative views, search techniques, history markers and hypertext links for navigation, animation, etc.
• PadPrints is a Pad++ application that can be used for web browsing. Research has shown this can be more effective than traditional browsers in a variety of common search tasks.

• Highly interactive nature of Pad++ yields rich interplay of cognitive processing, activity structure and dynamic representational changes.
• Better understanding of the coupling between interface components and cognition may help explain why zoomable multiscale interfaces seem so compelling and assist in effective design of alternative multiscale representations.
• How users manipulate icons, objects etc in Pad++ is part of their thinking process. They leave certain portals open to remind them of potentially useful information, they shift size to emphasize relative importance, move things around as items become more pressing.
• Studies of planning have typically focused on temporal order, but spatial is also important.
• We are constantly organizing and reorganizing our workplace to enhance performance. When space is used well, it reduces the time and memory demands of our tasks.
• Three functions of space:
  1. Arrangements that simplify choice. Examples: cover a hot handle so that touching it is not a choice. Lay out parts to be assembled in order.
  2. Arrangements that simplify perception. Place small parts of bicycle on sheet of paper, group similar pieces of jigsaw puzzle.
  3. Spatial dynamics that simplify internal computation. Strike a bundle of spaghetti on the table to “compute” the tallest one.

How would Distributed Cognition apply to an IDE? Think about features you’ve seen and/or features that seem helpful.