David Kirsh

We are quickly passing through the historical moment when people work in front of a single computer, dominated by a small CRT and focused on tasks involving only local information. Networked computers are becoming ubiquitous and are playing increasingly significant roles in our lives and in the basic infrastructure of science, business, and social interaction. For human-computer interaction o advance in the new millennium we need to better understand the emerging dynamic of interaction in which the focus task is no longer confined to the desktop but reaches into a complex networked world of information and computer-mediated interactions. We think the theory of distributed cognition has a special role to play in understanding interactions between people and technologies, for its focus has always been on whole environments: what we really do in them and how we coordinate our activity in them. Distributed cognition provides a radical reorientation of how to think about designing and supporting human-computer interaction. As a theory it is specifically tailored to understanding interactions among people and technologies. In this article we propose distributed cognition as a new foundation for human-computer interaction, sketch an integrated research framework, and use selections from our earlier work to suggest how this framework can provide new opportunities in the design of digital work materials.

This article addresses three main questions: What causes cognitive overload in the workplace? What analytical framework should be used to understand how agents interact with their work environments? How can environments be restructured to improve the cognitive workflow of agents? Four primary causes of overload are identified: too much tasking and interruption, and inadequate workplace infrastructure to help reduce the need for planning, monitoring, reminding, reclassifying information, etc… The first step in reducing the cognitive impact of these causes is to enrich classical frameworks for understanding work environments, such as Newell and Simon’s notion of a task environment, by recognizing that our actual workplace is a superposition of many specific environments – activity spaces – which we slip between. Each has its own cost structure arising from the tools and resources available, including the cognitive strategies and interpretational frameworks of individual agents. These cognitive factors are significant, affecting how easy or difficult it is to perform an action, such as finding a specific paper in a “mess” desk. A few simple examples show how work environments can be redesigned and how restructuring can alter the cost structure of activity spaces.

The theory of embodied cognition can provide HCI practitioners and theorists with new ideas about interac-tion and new principles for better designs. I support this claim with four ideas about cognition: (1) interacting with tools changes the way we think and perceive – tools, when manipulated, are soon absorbed into the body schema, and this absorption leads to fundamental changes in the way we perceive and conceive of our environments; (2) we think with our bodies not just with our brains; (3) we know more by doing than by seeing – there are times when physically performing an activity is better than watching someone else perform the activity, even though our motor resonance system fires strongly during other person observa-tion; (4) there are times when we literally think with things. These four ideas have major implications for interaction design, especially the design of tangible, physical, context aware, and telepresence systems.

To explore the question of physical thinking – using the body as an instrument of cognition – we collected extensive video and interview data on the creative process of a noted choreographer and his company as they made a new dance. A striking case of physical thinking is found in the phenomenon of marking. Marking refers to dancing a phrase in a less than complete manner. Dancers mark to save energy. But they also mark to explore the tempo of a phrase, or its movement sequence, or the intention behind it. Because of its representational nature, marking can serve as a vehicle for thought. Importantly, this vehicle is less complex than the version of the same phrase danced ‘full-out’. After providing evidence for distinguishing different types of marking, three ways of understanding marking as a form of thought are considered: marking as a gestural language for encoding aspects of a target movement, marking as a method of priming neural systems involved in the target movement, and marking as a method for improving the precision of mentally projecting aspects of the target.

In the course of daily life we solve problems often enough that there is a special term to characterize the activity and the right to expect a scientific theory to explain its dynamics. The classical view in psychology is that to solve a problem a subject must frame it by creating an internal representation of the problem’s structure, usually called a problem space. This space is an internally generable representation that is mathematically identical to a graph structure with nodes and links. The nodes can be annotated with useful information, and the whole representation can be distributed over internal and external structures such as symbolic notations on paper or diagrams. If the representation is distributed across internal and external structures the subject must be able to keep track of activity in the distributed structure. Problem solving proceeds as the subject works from an initial state in mentally supported space, actively constructing possible solution paths, evaluating them and heuristically choosing the best. Control of this exploratory process is not well understood, as it is not always systematic, but various heuristic search algorithms have been proposed and some experimental support has been provided for them.

We present data and argument to show that in Tetris - a real-time interactive video game - certain cognitive and perceptual problems are more quickly, easily, and reliably solved by performing actions in the world rather than by performing computational actions in the head alone. We have found that some translations and rotations are best understood as using the world to improve cognition. These actions are not used to implement a plan, or to implement a reaction; they are used to change the world in order to simplify the problem-solving task. Thus, we distinguish pragmatic actions ñ actions performed to bring one physically closer to a goal - from epistemic actions - actions performed to uncover information that is hidden or hard to compute mentally. To illustrate the need for epistemic actions, we first develop a standard information-processing model of Tetris-cognition, and show that it cannot explain performance data from human players of the game - even when we relax the assumption of fully sequential processing. Standard models disregard many actions taken by players because they appear unmotivated or superfluous. However, we describe many such actions that are actually taken by players that are far from superfluous, and that play valuable roles in improving human performance. We argue that traditional accounts are limited because they regard action as having a single function: to change the world. By recognizing a second function of action - an epistemic function - we can explain many of the actions that a traditional model cannot. Although, our argument is supported by numerous examples specifically from Tetris, we outline how the one category of epistemic action can be incorporated into theories of action more generally.

In dance, there is a practice called ‘marking’. When dancers mark, they execute a dance phrase in a simplified, schematic or abstracted form. Based on our interviews with professional dancers in the classical, modern, and contemporary traditions, it is fair to assume that most dancers mark in the normal course of rehearsal and practice. When marking, dancers use their body-in-motion to represent some aspect of the full-out phrase they are thinking about. Their stated reason for marking is that it saves energy, avoids strenuous movement such as jumps, and sometimes it facilitates review of specific aspects of a phrase, such as tempo, movement sequence, or intention, all without the mental and physical complexity involved in creating a phrase full-out. It facilitates real-time reflection.

Metacognition is associated with planning, monitoring, evaluating and repairing performance Designers of elearning systems can improve the quality of their environments by explicitly structuring the visual and interactive display of learning contexts to facilitate metacognition. Typically page layout, navigational appearance, visual and interactivity design are not viewed as major factors in metacognition. This is because metacognition tends to be interpreted as a process in the head, rather than an interactive one. It is argued here, that cognition and metacognition are part of a continuum and that both are highly interactive. The tenets of this view are explained by reviewing some of the core assumptions of the situated and distribute approach to cognition and then further elaborated by exploring the notions of active vision, visual complexity, affordance landscape and cue structure. The way visual cues are structured and the way interaction is designed can make an important difference in the ease and effectiveness of cognition and metacognition. Documents that make effective use of markers such as headings, callouts, italics can improve students’ ability to comprehend documents and ‘plan’ the way they review and process content. Interaction can be designed to improve ‘the proximal zone of planning’ – the look ahead and apprehension of what is nearby in activity space that facilitates decisions. This final concept is elaborated in a discussion of how e-newspapers combine effective visual and interactive design to enhance user control over their reading experience.

Contemporary choreography offers a window onto creative processes that rely on harnessing the power of sensory sys- tems. Dancers use their body as a thing to think with and their sensory systems as engines to simulate ideas non- propositionally. We report here on an initial analysis of data collected in a lengthy ethnographic study of the making of a dance by a major choreographer and show how translating between different sensory modalities can help dancers and choreographer to be more creative.

Why do people create extra representations to help them make sense of situations, diagrams, illustrations, instructions and problems? The obvious explanation— external representations save internal memory and com- putation—is only part of the story. I discuss seven ways external representations enhance cognitive power: they change the cost structure of the inferential landscape; they provide a structure that can serve as a shareable object of thought; they create persistent referents; they facilitate re- representation; they are often a more natural representation of structure than mental representations; they facilitate the computation of more explicit encoding of information; they enable the construction of arbitrarily complex structure; and they lower the cost of controlling thought—they help coordinate thought. Jointly, these functions allow people to think more powerfully with external representations than without. They allow us to think the previously unthinkable.

This paper examines some of the methods animals and humans have of adapting their environment. Because there are limits on how many different tasks a creature can be designed to do well in, creatures with the capacity to redesign their environments have an adaptive advantage over those who can only passively adapt to existing environmental structures. To clarify environmental redesign I rely on the formal notion of a task environment as a directed graph where the nodes are states and the links are actions. One natural form of redesign is to change the topology of this graph structure so as to increase the likelihood of task success or to reduce its expected cost, measured in physical terms. This may be done by eliminating initial states hence eliminating choice points; by changing the action repertoire; by changing the consequence function; and lastly, by adding choice points. Another major method for adapting the environment is to change its cognitive congeniality. Such changes leave the state space formally intact but reduce the number and cost of mental operations needed for task success; they reliably increase the speed, accuracy or robustness of performance. The last section of the paper describes several of these epistemic or complementary actions found in human performance.

Why do people create extra representations to help them make sense of situations, diagrams, illustrations, instructions and problems? The obvious explanation – external representations save internal memory and computation – is only part of the story. I discuss eight ways external representations enhance cognitive power: they provide a structure that can serve as a shareable object of thought; they create persistent referents; they change the cost structure of the inferential landscape; they facilitate re-representation; they are often a more natural representation of structure than mental representations; they facilitate the computation of more explicit encoding of information; they enable the construction of arbitrarily complex structure; and they lower the cost of controlling thought – they help coordinate thought.

The question of how to conceive and represent the context of work is explored from the theoretical perspective of distributed cognition. It is argued that to understand the office work context we need to go beyond tracking superficial physical attributes such as who or what is where and when and consider the state of digital resources, people’s concepts, task state, social relations, and the local work culture, to name a few. In analyzing an office more deeply, three concepts are especially helpful: entry points, action landscapes, and coordinating mechanisms. An entry point is a structure or cue that represents an invitation to enter an information space or office task. An activity landscape is part mental construct and part physical; it is the space users interactively construct out of the resources they find when trying to accomplish a task. A coordinating mechanism is an artifact, such as a schedule or clock, or an environmental structure such as the layout of papers to be signed, which helps a user manage the complexity of his task. Using these three concepts we can abstract away from many of the surface attributes of work context and define the deep structure of a setting—the invariant structure that many office settings share. A long-term challenge for context-aware computing is to operationalize these analytic concepts.

It is sometimes argued that if PDP networks can be trained to make correct judgements of grammaticality we have an existence proof that there is enough information in the stimulus to permit learning grammar by inductive means alone. This seems inconsistent superficially with Gold's theorem and at a deeper level with the fact that networks are designed on the basis of assumptions about the domain of the function to be learned. To clarify the issue I consider what we should learn from Gold's theorem, then go on to inquire into what it means to say that knowledge is domain specific. I first try sharpening the intuitive notion of domain specific knowledge by reviewing the alleged difference between processing limitations due to shortage of resources vs shortages of knowledge. After rejecting different formulations of this idea, I suggest that a model is language specific if it transparently refer to entities and facts about language as opposed to entities and facts of more general mathematical domains. This is a useful but not necessary condition. I then suggest that a theory is domain specific if it belongs to a model family which is attuned in a law-like way to domain regularities. This leads to a comparison of PDP and parameter setting models of language learning. I conclude with a novel version of the poverty of stimulus argument.

A complementary strategy can be defined as any organizing activity which recruits external elements to reduce cognitive loads. Typical organizing activities include pointing, arranging the position and orientation of nearby objects, writing things down, manipulating counters, rulers or other artifacts that can encode the state of a process or simplify perception. To illustrate the idea of a complementary strategy, a simple experiment was performed in which subjects were asked to determine the dollar value of collections of coins. In the no-hands condition, subjects were not allowed to touch the coin images or to move their hands in any way. In the hands condition, they were allowed to use their hands and fingers however they liked. Significant improvements in time and number of errors were observed when S's used their hands over when they did not. To explain these facts, a brief account of some commonly observed complementary strategies is presented, and an account of their potential benefits to perception, memory and attention.

The type of principles which cognitive engineers need to design better work environments are principles which explain interactivity and distributed cognition: how human agents interact with themselves and others, their work spaces, and the resources and constraints that populate those spaces. A first step in developing these principles is to clarify the fundamental concepts of environment, coordination, and behavioural function. Using simple examples, I review changes the distributed perspective forces on these basic notions.

The objective of this essay is to provide the beginning of a principled classification of some of the ways space is intelligently used. Studies of planning have typically focused on the temporal ordering of action, leaving as unaddressed questions of where to lay down instruments, ingredients, work-in-progress, and the like. But, in having a body, we are spatially located creatures: we must always be facing some direction, have only certain objects in view, be within reach of certain others. How we manage the spatial arrangement of items around us is not an afterthought: it is an integral part of the way we think, plan, and behave. The proposed classification has three main categories: spatial arrangements that simplify choice; spatial arrangements that simplify perception; and spatial dynamics that simplify internal computation. The data for such a classification is drawn from videos of cooking, assembly and packing, everyday observations in supermarkets, workshops and playrooms, and experimental studies of subjects playing Tetris, the computer game. This study, therefore, focuses on interactive processes in the medium and short term: on how agents set up their workplace for particular tasks, and how they continuously manage that workplace.

Architecture is about to enter its first magical phase: a time when buildings actively co-operate with their inhabitants; when objects know what they are, where they are, what is near them; when social and physical space lose their type coupling; when wall and partitions change with mood and task. As engineers and scientists explore how to digitse the world around us, the classical constraints of design, ruled so long by the physics of space, time, and materials, are starting to crumble. Documents can be laid down in one place, automatically cloned, and a copy picked up in another. Meetings scheduled for 9am to 10am can be joined by latecomers at noon, who then participate in a captured form of the event and are ‘edited into’ the past. People on the West coast of the USA can participate, in a telepresent way, with their colleagues on the East coast, and hold a meeting against a virtual backdrop, such as a production line in their Taiwanese factory. Walls seem to dematerialise, remote objects can be touched virtually, shaped, passed through one another. Technology is moving inexorably so that being in one place at a time no longer need dominate how we work and play. Material boundaries are losing their meaning, and interface and information space are catch words that architects must master. In this article I will discuss some of the theoretical ideas shaping our new conception of form, function, and interactivity. My view is that of a cognitive scientist interested in how cognition is distributed throughout our environment. Since the ground rules defining the structure of environments are changing, our very idea of how we are embedded in the world is changing. Architecture is at a new frontier.

Multimedia technology offers instructional designers an unprecedented opportunity to create richly interactive learning environments. With greater design freedom comes complexity. The standard answer to the problems of too much choice, disorientation, and complex navigation is thought to lie in the way we design interactivity in a system. Unfortunately, the theory of interactivity is at an early state of development. After critiquing the decision cycle model of interaction—the received theory in human computer interaction—I present arguments and observational data to show that humans have several ways of interacting with their environments, which resist accommodation in the decision cycle model. These additional ways of interacting include: preparing the environment, maintaining the environment, and reshaping the cognitive congeniality of the environment. Understanding how these actions simplify the computational complexity of our mental processes is the first step in designing the right sort of resources and scaffolding necessary for tractable learner controlled learning environments.

We present data and argument to show that in Tetris—a real-time interactive video game—certain cognitive and perceptual problems are more quickly, easily, and reliably solved by performing actions in the world rather than by performing computational actions in the head alone. We have found that some translations and rotations are best understood as being used to implement a plan, or to implement a reaction. To substantiate our position we have implemented a computational laboratory that lets us record keystrokes and game situations, as well as allows us to dynamically create situations. Using the data of over 30 subjects playing 6 games, tachistoscopic tests of some of these subjects, and results from our own successful efforts at building expert systems to play Tetris, we show why knowing how to use one’s environment to enhance speed and robustness are important components in skilled play.

When people make sense of situations, illustrations, instructions and problems they do more than just think with their heads. They gesture, talk, point, annotate, make notes and so on. What extra do they get from interacting with their environment in this way? To study this fundamental problem, I looked at how people project structure onto geometric drawings, visual proofs, and games like tic tac toe. Two experiments were run to learn more about projection. Projection is a special capacity, similar to perception, but less tied to what is in the environment. Projection, unlike pure imagery, requires external structure to anchor it, but it adds ‘mental’ structure to the external scene much like an augmented reality system adds structure to an outside scene. A person projects when they look at a chessboard and can see where a knight may be moved. Because of the cognitive costs of sustaining and extending projection, humans make some of their projections real. They create structure externally. They move the piece, they talk, point, notate, represent. Much of our interactivity during sense making and problem solving involves a cycle of projecting then creating structure.

We designed an experiment to explore the learning effectiveness of three different ways of practicing dance movements. To our surprise we found that partial modeling, called marking in the dance world, is a better method than practicing the complete phrase, called practicing full-out; and both marking and full-out are better methods than practicing by repeated mental simulation. We suggest that marking is a form of practicing a dance phrase aspect-by-aspect. Our results also suggest that prior work on learning by observation and learning by mental practice may not scale up to complex movements.

Can virtual engagement enable the sort of interactive coupling with objects enjoyed by archaeologists who are physically present at a site? To explore this question I consider three points: 1) Tangible interaction: What role does encounter by muscle and sinew play in experiencing and understanding objects? 2) Thinking with things. What sorts of interactions are involved when we manipulate things to facilitate thought? 3) Projection and imagination. Archaeological inquiry involves processes beyond perception. Material engagement of things stimulates these processes. What must be present in a virtual environment to recreate the feel of material engagement? I conclude that nothing, in principle, prevents future virtual environments from supporting this material engagement of digital versions of artifacts. But, there is much that remains to be understood about how to realize this material engagement, both at a technological and a cognitive level.

In the scientific study of mind a distinction is drawn between explicit knowledge— knowledge that can be elicited from a subject by suitable inquiry or prompting, can be brought to consciousness, and externally expressed in words—and implicit knowledge—knowledge that cannot be elicited, cannot be made directly conscious, and can- not be articulated. Michael Polanyi (1967) argued that we usually ‘know more than we can say’. The part we can articulate is explicitly known; the part we cannot is implicit.

Virtual environment landmarks are essential in wayfinding: they anchor routes through a region and provide memorable destinations to return to later. Current virtual environment browsers provide user interface menus that characterize available travel destinations via landmark textual descriptions or thumbnail images. Such characterizations lack the depth cues and context needed to reliably recognize 3D landmarks. This paper introduces a new user interface affordance that captures a 3D representation of a virtual environment landmark into a 3D thumbnail, called a worldlet. Each worldlet is a miniature virtual world fragment that may be interactively viewed in 3D, enabling a traveler to gain first-person experience with a travel destination. In a pilot student conducted to compare textual, image, and worldlet landmark representations within a wayfinding task, worldlet use significantly reduced the overall travel time and distance traversed, virtually eliminating unnecessary backtracking.

An experiment was performed to test the hypothesis that people sometimes take physical actions to make themselves more effective problem solvers. The task was to generate all possible words that could be formed from seven Scrabble letters. In one condition, participants could use their hands to manipulate the letters, and in another condition, they could not. Results show that more words were generated with physical manipulation than without. However, an interaction was obtained between the physical manipulation conditions and the specific letter sets chosen, indicating that physical manipulation helps more for generating words in some circumstances than in others. Overall, our findings can be explained in terms of an interactive search process in which external, physical activity effectively complements internal, cognitive activity. Within this framework, the interaction can be explained in terms of the relative difficulty of generating words from the letters given in the different sets.