Engaging Space through Embodiment

This concept of space, based on our perception and cognition, underlies the design of information environments. The electronic emulation of space – here called anthropic cyberspace – extends beyond our material world to include memory, symbol and metaphor. It is the matrix of human experience.

Designing anthropic cyberspace harnesses our use of space to think, navigate, interact. Anthropic cyberspace is based on our spatial construct – the mental image of our surroundings. It puts the us at ground-zero, presenting us with primary views of an information field. Up and down, front and back are implicit in such views, grounding and orienting us. Directed lighting, shadows and artificial horizons evoke their familiar counterparts in nature. Surprisingly, these all depend on the tacit presence of the user's body, for it is the body that stands at the origin of perceived space. Our primary senses, oriented frontally, distinguish back from front, left from right. Our natural posture establishes verticality, up and down. Even the flat horizon – always at eye level – is determined by the body.

It may seem strange that our bodies should matter so in on-line environments. Cyberspace is, after all, one of our most abstract artifacts – comprised of symbols and electronic signals. But embodiment is a pragmatic way for us to engage the information of cyberspace. It capitalizes on our natural skills, extending them to the management of abstraction and conceptualization. Designers of cyberspace ignore this incorporation at their peril.


Information and Movement

We interact with space through sense and bodily motion. And it is motion that reveals the subtle influence of the body on the shape of cyberspace. Cyberspace, like conventional space, is shaped by our means of navigation. We need only look at city plans over the centuries to see how movement has shaped their development. The shape of a city founded on river routes, for instance, is quite different from that of one based on the automobile. We will see navigation’s influence on non-physical communities as well.

But before going further, we should examine movement from a sensory and cognitive standpoint. What is motion? How does it relate to information? How is motion relevant to symbolic environments like the Internet and cyberspace communities?

Motion enhances our bodily experience. We explore to seek things otherwise hidden from view. After all, our eyes can’t see around corners. Our senses depend on the relative motion between the body and its ambience. Air conveys scent and sound as light does vision. We know that touch, too, depends on on motion as our hands brush the surface of stone, feeling its rough texture.
 

The Cognitive Nature of Dynamic Motion

We are accustomed to moving dynamically through space. Dynamism is fluid motion, sequential moments stitched together in our minds. Fleeting glances and sounds comprise the elements of dynamic motion. Zeno's Paradox implies that motion cannot be comprised of units. If so, an arrow shot from a bow would never reach its mark, having first to travel half the distance and prior to that 1/4 the distance, 1/8 and so on. But whether motion – or time itself – is a sum of instances is not at issue. What matters is that our experience of motion is mentally derived from finite sensations. For motion, like space, is a byproduct of our perception and cognition. It is a way we explain change to ourselves – filling the gaps between events.

We engage in this cognitive form of motion continually. Take memory for instance. You may be seated in a room as you read this text. Close your eyes and recall the events of your day – waking, preparing for work, coming to this place. Imagine yourself entering the room, reaching for this paper and sitting down. You have just engaged in two forms of nonphysical movements. The first type, rising from your bed and walking into rooms, were accomplished without moving from your seat. The actions mimicked dynamic motion though they were imaginary.


Categorical Motion and Electronic Media

The second form of motion was more subtle. Mentally you moved back in time to your bed. It was a categorical shift, an instant relocation. Compared with fluid dynamism, categorical motion takes us abruptly to our destination. It is more of a change of scene than a physical motion.

The physical equivalent of categorical motion is hard to visualize. However, it is familiar to us from fiction and fairy tales. Time travel, appearance and disappearance, and teleportation magically transfer us to new places and times. Categorical motion in fiction is a metaphor for our shifts of attention.

Both dynamic and categorical motion are present in electronic environments. While dynamism is a relative newcomer, categorical motion has long been a favorite in cyberspace. However accessing files, going to Web sites, and entering chat rooms lack the richness of moving dynamically. The motion is categorical, instantaneous and disembodied. You simply arrive there.

Surfing the Web is the electronic equivalent of cognitive motion. The difference is that it occurs outside the mind – someone looking over your shoulder can follow your thoughts by watching the screen. This is most apparent with motion encountered in Multi-user domains and graphic Worlds. A second difference – often overlooked – is the doubling of the information source. With cognitive motion, the thinker, the thought and motion are unique. With electronic media, however, the source is always duplicated in transmission. This is true of television, radio or even the telephone.

Though we are not present to the newscaster on the evening news, he is to us. However, if a medium were interactive then we, too, would be duplicated through an electronic double. Take a simple telephone conversation. We never actually hear each other on the phone. We hear instead replications of our voices that give us the illusion of proximity. Space seems to have folded – we haven't moved, yet we have moved.

Episodic motion

One type of categorical motion, episodic motion, is presented in glimpses – sequential views taken along a path. This is the motion we encounter in some computer games, notably Myst and Riven. High-resolution, still images effectively reside in a spatial matrix. The user's sense of motion is a product of categorical jumps between these linked views. Selecting the top of a mountain in one scene, for instance, results in a categorical leap to its summit in the next.

Episodic motion assumes a spatial narrative created from glances along a route. It works on-line because we navigate this way conventionally – our flashing glances comprise dynamic motion. Episodic motion requires that consecutive scenes have enough in common that they can be spatially related. In fact, sequences of images that differ slightly can simulate dynamic motion if viewed in quick succession. We see this in film and video whose spatial consistency conveys an apparent yet fluid motion.

Episodic motion is common in text-based Multi-user Domains (MUDs) in which users navigate between rooms. Text-MUDs resemble Internet Relay Channel (IRC) chat rooms in that users communicate live with each other. In both cases, MUDs and IRCs,

rooms are on-line locales for action and conversation. However, MUDs are much more spatially elaborate – each containing hundreds of rooms.

MUD rooms are discrete locations within an overall spatial matrix. Since there are no images in text domains, each room is defined by a short paragraph describing, say, a room in a castle, the deck of the spaceship, or a grand ballroom. Often as not, the description includes commands for likely exit directions or destinations. Typing these moves the player from place to place.


Embodiment in Text-Domains

Actually the player herself doesn't move – her representation does. Networked computing, like any other electronic medium, doubles the source. For the player to interact with other inhabitants inthe MUD, she has to be doubled, represented by an avatar. The avatar is her online embodiment.

Movement, dialog and actions are enacted vicariously by this double. Despite the user’s embodiment movement between spaces is abstract, categorical. Orientation and bodily references apply only at the room destinations, not in the passage between them. Traveling between rooms is instantaneous despite cues offered in the room descriptions. While to others an avatar may materialize magically or vanish in a puff of smoke, the avatar’s user only experiences an instant transition between spaces. The only link to dynamic movement is that these spaces are described as part of an underlying spatial matrix.
 

Static motion

A third type of movement employs the symbolic nature of media, transposing us without reference to dynamic motion or materiality. This is static motion – the movement between states. This last, most abstract, form of categorical movement lets us go from one representation of an object to another. While the object does not change, its representation does. Here movement is more of a metaphor for shifts of attention, or even the learning process itself. The gradual recognition of an object, for instance, takes us from a sensory mode to one of cognition and memory. Neither viewer nor the object changes, but the subject's state in the our mind does.


In media these transformations are more apparent and public. For example, one of the pleasures of children's books is the mutual reinforcement of text and illustration. Our attention flickers from word to image – each distinct states of the story. Moving between the representations rounds-out the narrative. Readers familiar with computer graphics – particularly those of computer-aided design – know that users can select among graphic and text modes of representation. The same file can be manifested as a two or three-dimensional image, a wire frame or rendering, a text or table – even an animation – each state having its own virtue and use. No change in user or file, only in the states of presentation and mind.

Static movement is common also on the Internet. Internet browsers also offer us Web sites in graphic or text modes. Likewise, computer programmers can go behind the scenes to manipulate graphic software with programming code. This last example is pervasive in Multi-user domains and MOOs (MUD, Object-Oriented). Users' commands in MUDs are divided between those that effect avatar actions and those that affect the MUD's program. Metaphorically, the former take place onstage while the latter occur backstage. Going offstage entails static motion and is managed with several commands starting with @: @dig, @look, @sweep, @paranoid.

In using these commands, the user momentarily drops the avatar mask to engage with the MUD's operating system. Unlike episodic motion that maintains the illusion of fluid dynamism, the categorical motion evoked by @go is static since the player has to leave the state of play, “move” by means of the operating system, and re-enter the domain at another location. Similarly, logging in and out of the domain also requires static movement as we enter or leave the state of play. Teleportation requires static movement to effect a discrete motion, moving players to other points in the MUD.
 

Reprise

This would be a good point to summarize the forms of movement describe so far. Essentially there are two classes of motion: dynamic and categorical. Dynamic motion may be physical but it can also be simulated in film or animation. It is fluid and consistent – we are unable to distinguish its elements. Conversely, categorical movement is fragmentary – a series of images whose consistency determines whether the categorical motion is episodic, discrete or static.

Of these three categorical motions, episodic most resembles dynamic motion, its components consistent and sequential, adhering to an underlying matrix. It is categorical because of its abrupt relocation. Yet it maintains its incremental spatial consistency. Conversely, discrete motion flashes us to our destination, regardless of its place or time.

However when we arrive there – whether episodically or discretely – we engage the new scenario dynamically. With these motions, destinations abide by conventions of material reality – we are embodied there regardless of our mode of transport. Of the three types of categorical motion only static motion takes us nowhere. Or – more precisely – takes us to no new place. With it we move between states of the place and moment we occupy.

A brief example will help to illustrate the relationship between these types of movement. Imagine going to a film at the theater. As we watch the characters move we experience dynamic motion – not in the sense of moving physically but as an impression made by the film. We cannot distinguish units of time less than 1/30th of a second. For this reason the fast sequence of still shots appears to be a single, moving image. Despite our dynamic experience, film is in fact a high-speed presentation of categorical motion – specifically, episodic motion.

If we removed every other cell in the film we would notice the incremental nature of the film. By removing two or three sequential cells, movement becomes choppy, though we can still make out the movie’s spatial structure. For example, a person walking towards a house eventually gets to the door and enters. Enough intervening scenes assure us that the character walked there, encountering steps and an angry dog along the way. The abrupt changes between images are not enough to break the episodic narrative.

If we continue to edit – removing, say, 19 of 20 sequential images – we begin to experience discrete images. The character entering the house in the original version now stands on the sidewalk one moment and in a living room the next. Only viewers familiar with the original film would know what happened and only to them would the movement be episodic. For most of us, though, the motion seems discrete since we no longer can follow the spatial narrative. Seeing the entire film this way would be disorienting and a bit unnerving. We need more to orient us than the images alone provide.

Film actually has a specialized visual language to deal with narrative breaks. After all, hardly a movie was shot in a single camera take. Since early in film’s history, viewers have become accustomed to cuts, fades, pans and a host of other techniques to follow the action. All these tacitly move the viewer through the film.
 

Vicarious motion and domain structure

We can learn a lot about a domain’s structure by looking at the movement used to navigate it. While in many graphic domains and Worlds motion is animated and dynamic, users of text-based MUDs move episodically between spaces. This movement can be tedious for players who know their destination and prefer not to go through the intervening spaces. And such impatience betrays a familiarity with the domain, for new users prefer to explore the MUD episodically. It lets them learn the relationships of spaces with which they build a cognitive map of the domain.

Many text domains have a core community of rooms that are arrayed spatially, using cardinal directions to orient the user. These spaces may have been built originallly by the MUD administrators or constructed later by citizens of the domain, depending on the MUD's operation. Invariably, one of these rooms is an entryway into the domain. Other spaces are likely to include gathering places for users' avatars, rooms descriptive of interiors or landscapes, and transit stations from which users may take a bus, taxi, or helicopter to other parts of the MUD.

MUD players arrive at the domain categorically by using a browser or other Internet application. Once there, they log on, assume an avatar and enter the MUD. This entry is accomplished statically as players go from the operational state of the MUD to its spatial emulation. Upon entry, users are embodied through their avatars. Actions within the rooms are described dynamically, much as one might read them in a novel. Avatars engage in real-time dialog, wave to one another, and examine objects in the room.

But the artifice is exposed in the room's description and its "logical exits". By using these exits players move episodically from one room to the next. Though this motion is called "walking" it is more like popping abruptly into adjacent rooms. This categorical movement is unique to text-MUDs. Graphic domains and Worlds use animation to dynamically move the avatars about. And since the spatial reference is literal in a world like Alphaworld, users are not restricted in their movements. They may move dynamically to any point or in any direction. Whereas the spatial matrix of the text-MUD may be described as a flowchart of options as an LAN, that of a World is better relayed as a conventional map or plan

Dynamic simulation – whether animated or episodic – takes time in both text or graphic MUDs. This is even more so for episodic motion. For an avatar to go to another room, the user refers to the current room's exit options, types in a directional command, waits for the system to respond, reads the description of the ensuing space and its occupants. This gets tedious after the user is acquainted with the spaces and, especially, if he encounters no one in the rooms.
 

Vacancy and Discrete Motion

Consequently, many players congregate in the area around the entry. Going beyond a convenient walking radius will present new spaces but also a decreasing chance of encountering anyone there. As a result many new visitors to a MUD stay close to the entry, chat with those they encounter and leave. If they take one of the MUD's conveyances – say, a train – they may get off at any station, but their walking radius around the station will likely be smaller than that at the entry point. It appears that a majority of MUD spaces are unoccupied. And there are a few good reasons why this is so.

Though the initial construction of the MUDs rooms was done by the administrators, many MUDs and MOOs allow their citizens to create their own spaces. As noted previously, these spaces may or may not be connected to the original spatial matrix. Regardless, users are free to create one or more spaces. Given that users aren't present in the MUD at all times it follows that the majority of spaces will be unoccupied. This overbuilding is aggravated by the turnover of citizens and a reluctance of administrators to destroy the work of the MUD community.
In building a new space in a text-domain or MOO the user must separately program the entry and exits from the room. This is a simple procedure if the builder connects to spaces she has already built. But if she wants to link to another user's construction she first must obtain permission to do so. For example, should Mary wish to build to the west of Tom's room, she has to first get his permission to add an exit/entry to Tom's space. This would allow visitors move episodically to her room by going west.

This is well and good if Mary can get Tom's okay to remodel his space. But Tom may not be cooperative or available, and so the two rooms remain isolated from one another. Although avatars can’t move episodically between the spaces, they can teleport since each space has a specific address. By using the @go command with the address, an avatar can move discretely between these rooms.

But discrete motion exacerbates problems already caused by overbuilding. Since dedicated MUDders spend as much of their time building as chatting, new construction often occurs outside the spatial matrix. This requires users to a) teleport in order to access them and b) know the address of the space. The streets of MUDs are empty because many users have to teleport to their rooms. To the casual MUD user these spaces are invisible – as are their occupants.

While discrete motion is convenient – and sometimes necessary – it taxes the social activity of the domain. Users do not interact with colleagues that they would otherwise meet. Chance encounters are minimized when travel is determined solely by destination rather than the path. The more abstract the motion, the less spatial it is – and the less conducive it is to social interaction.


The Spatial Products of Dynamic and Categorical Motion

Motion and user embodiment have a noticeable effect on the shape of online communities. In the diagram (fig. 4) I have shown a generic MUD with features that will be described below and referenced by number. The logical adjacency map of a text-MUD often reveals an orthogonal core structure in which room locations are subject to an overarching spatial matrix (1). These rooms are accessed directionally by going north, east, south, or west and may be mapped with respect to one another. Movement among these rooms is largely episodic.

The core structure includes the MUD entry (2) often in the form of a lobby, gateway, or even a bedroom closet. Though text-MUD entries may be exited episodically or discretely, they are nearly always entered statically – arriving from the MUDs operational login screen. The entry appears as a simple node on a logical adjacency map but it is the point of origin for all the MUDs social activity. This is a product of channeled, static motion: population density is often highest around the domain's entry. (3)

As one leaves the vicinity of the MUD entry, the population density drops off quickly. However, remote locations (4) – such as bars or nightclubs – may hold high densities of avatars and the immediate surrounding area benefits from the activity. Users familiar with the MUD prefer to teleport there directly from the entry rather than wading episodically through the intervening rooms.
Unlike episodic motion, discrete motion does not create cross-traffic. For example, a shopping mall modeled on dynamic or episodic motion benefits from the polar disposition of anchor stores. As shoppers travel between them they encounter many smaller shops. If, instead, users moved discretely between anchor stores the little shops would swiftly go out of business. In a physical community we expect development between a point of entry and a social watering-hole. But the discrete motion of text-MUDs isolates these poles, rendering each a concentric hub for social activity. (5)

The eight-pointed star of Alphaworld

The effect of different kinds of motion may also be observed in graphic domains and Worlds. Alphaworld presents a striking example of how an online community is shaped by its users' movements. Unlike its text-based cousins, a graphic domain like Alphaworld employs dynamic motion through the animation of its avatars. There is no need for the incremental, directional movement found in text-MUDs since the spatial matrix is obvious and avatars move freely about it.

Despite its graphic three-dimensional presentation, Alphaworld shares many features with text-domains. It, too, has a point of entry near the origin of the spatial matrix. Dynamic motion prevails within a convenient walking radius from the entry. Construction is dense within this radius and it is apparent that the community grew radially over time. (fig. 5) An orthogonal grid of streets defines the layout, though their dense mesh loosens as one moves toward its perimeter. Green space and larger city blocks appear toward this edge.

The author wishes to credit the enthusiastic efforts of his students at the schools of architecture at the New Jersey Institute of Technology and University of Michigan in making this research possible.

 

Bibliography

Altman, Irwin. 1981. The environment and social behavior. New York: Irvington Publishers Inc.
Anders, Peter. 1996. “Envisioning cyberspace: The design of on-line communities,” in Design computation: Collaboration, reasoning, pedagogy. McIntosh, P. and F, Ozel, eds. Proceedings of ACADIA 1996 Conference, Tucson, Arizona. pp.55- 67.
Anders, Peter. 1999. Envisioning Cyberspace. New York: McGraw-Hill. pp. 79-98.
Benedikt, Michael. 1994. Cyberspace: First Steps. Cambridge, Mass.: MIT Press.
Bergson, Henri. 1988. Matter and memory. Cambridge, Mass.: MIT Press. pp.195, 202.
Bødker, Sussane. 1990. A human activity approach to user interface design. Hillsdale, N. J.: Lawrence Erlbaum Associates.
Bruckman, Amy, and Mitchel Resnick. 1995. “The MediaMOO project: Construc-
tionism and professional community.” Convergence 1(1).
Curtis, Pavel. 1992. “MUDding: Social phenomena in text-based virtual realities,” in proceedings of 1992 Conference on Directions and Implications of Advanced Computing. Berkeley, Calif., May 1992. XeroxParc technical report CSL- 92-4. Available by anonymous ftp: parcftp.parc.xerox.com in directory /pub/MOO/papers/ in files DIA92.
Curtis, Pavel, and David Nichols. 1993. “MUDs grow up: Social virtual reality in the real world” (Austin, TX) 1993. Available by anonymous ftp: parcftp.parc.xerox.com in directory /pub/MOO/papers/ in file MUDsGrowUp.
Gibson, James J. 1982. Reasons for realism: Selected essays of James J. Gibson. Reed, E., and R. Jones, eds. Hillsdale, N. J.: Lawrence Erlbaum Associates.
Heim, Michael. 1995. “The design of virtual reality,” in Cyberspace, cyberbodies, cyberpunk: Cultures of technological embodiment. Featherstone, Michael, and Roger Burrows, eds. London: Sage Publications. pp. 65-78.
Johnson-Laird, P.N. 1981. “Mental models in cognitive science,” in Perspectives on cognitive science. Norman, D. A., ed. Hillsdale, N.J.: Lawrence Erlbaum Associates.
Lakoff, George, and Mark Johnson. 1980. Metaphors we live by. Chicago: University of Chicago Press.
Laurel, Brenda. 1986. “Interface as mimesis,” in User centered system design. Norman, D. A., and S. W. Draper, eds. Hillsdale, N.J.: Lawrence Erlbaum Associates. pp. 67-85.
Piaget, Jean. 1960. The child's conception of physical causality. Paterson, NJ: Littlefield, Adams & Co.
Spoehr, Kathryn T. 1994. “Enhancing the acquisition of conceptual structures through hypermedia,” in Classroom Lessons. McGilly, K., ed. Cambridge, Mass.: The MIT Press.
Turkle, Sherry. 1996. Life on the screen. New York: Simon and Shuster.
Walter, Eugene V. 1988. Placeways: A theory of the human environment. Chapel Hill, N. C., and London: University of North Carolina Press.