This report is also available as an Acrobat file.

Contents
The Design of Virtual Environments with particular reference to VRML

Representation

The viewpointless view

All the projections we have discussed assume that a viewpoint is necessary, and indeed a view without a viewpoint seems a nonsense. However, children's drawings and some other artforms arguably provide just this: the objects they display carry their canonical views with them as properties of the objects themselves, so that human beings are always displayed full-face, animals are viewed in profile, rainbows are always visible throughout their length. In a limited sense, this is an approach with which computer users are familiar - they see it in an iconic graphical user interface.

Taking liberties with realism

There are a number of attributes of the perceived world which, rather than copying slavishly, we may be able to 're-use' for our own purposes, in order to clarify or enhance meaning. Examples include focus, distance, atmosphere and scale.

Focus

We described how the work of Cezanne attempted to capture the process of seeing. Others have used focus in similar ways. It seems certain that Rembrandt's self-portraits owe a part of their extraordinary presence to his depiction of the differential focus of the plane of the face, so that the eyes, the hypnotic subject for any painter observing his own face in a mirror, are in sharp focus and the tip of the nose and the distant parts of the head are relatively defocussed. Baxandall (1985 p80) shows how Chardin uses selectivity in the sharpness and softness of contours, to capture the way in which the eye might prefer certain trajectories across the scene. In so doing, Chardin almost certainly hopes to lead the viewer's eye on a similar journey in negotiating the parts of his painting. A crude example from this century would be the use of selective focussing to take in and out of attention different parts of a movie scene, perhaps 'pulling focus' from one character's face to another's.

Distance

Distance is generally perceived as an inconvenience: we have to travel from place to place at a certain speed. Taking advantage of the non-physical nature of virtual reality, we can all but abolish the lapse of time in translating ourselves from place to place. Yet time is an important carrier of information: we use it as meta-information to discern where data (for example sound) is coming from. In place of time-delays arising from the accidents of place, we could use time-delays as an analogue of other qualities, for example the remoteness of one idea (as embodied by an object, building, etc.) from another.

Atmosphere

The effect on clarity and colour of distant objects, is another accident of reality which imposes difficulties and yet which might be turned to advantage. In physical reality, atmosphere is a function of distance and climate: there is nothing to be gained by saddling ourselves with such accidents to no purpose. However we can find uses for the effect. For example, we could assist the user in giving attention to certain features by partially obscuring some and clarifying others. In making such a decision, we are again designing.

The uses of scale

If we took a narrow view of VR as photorealism in 3D, we would deny ourselves all the possibilities of using scale to create meaning. One has only to look at pre-Renaissance painting, or the films of most animators, to see what can be done once unthinking adherence to perspectivally 'correct' scale is abandoned.

Information spaces: database to Virtual Museum

Virtual environments may have potential as frameworks for knowledge, even where that knowledge is not inherently spatial. Pieces of information may be situated in a virtual space. The analogy with museums has led to the construction of virtual galleries, and certainly non-physical environments do seem to offer some possible benefits:

Collection of artefacts which used to be housed in one place can be reassembled, perhaps in a reconstruction of their original location. So for example a picture collection dispersed through warfare or natural disaster can be partly reconstituted in a virtual sense.
Extant collections can be made available in a form which resembles their actual setting to those unable to have access to the real thing.
Artefacts can be placed in new contexts evoking their original use or setting (something many physical museums struggle to reconcile with scholarship, curatorial care and so forth).
The emphasis on static, preservable objects associated with most physical museums can be overthrown. Sound and other temporal or fleeting phenomena can be included just as well as physical artefacts.
New kinds of artefacts can be created to take advantage of the special characteristics of such environments.

However, as always we must beware of any naive assumption that the design of virtual museums is a matter of replicating the style and character of the physical museums with which we are familiar. Three virtual dimensions may be no better than two in representing the structures of knowledge.

Some of the difficulties of virtual galleries are:

Objects can usefully be considered to occupy a place in as many dimensions as there are metrics for describing them (in database terminology, the space has as many dimensions as there are fields in the record). So for example a painting whose date, author, material, place of origin and main subject are recorded is already situated in a five-dimensional space. Three dimensions are not then enough to do justice to the subtlety of the relationship between each artefact and its neighbours. Of course those dimensions may be reduced to a three-dimensional space for the purposes of display (just as they are in a physical gallery when paintings are exhibited chronologically for example). But this raises another important objection:
We are faced with a difficult choice. We could place objects in the 3-space we design, and not allow the user to reconfigure the space. In so doing we transfer a major disadvantage of the physical gallery to the virtual one: objects can no more be viewed in new contexts than they could in a physical space. Alternatively, we could allow users to continually re-order the 'gallery' according to the needs of the moment (for example, on one occasion to see all Italian paintings together, on another, all from the 18th Century, on another all those painted using tempera). Now the risk is that we lose key advantages of virtual environments over conventional databases, particularly the easy memorability of spatial organisation: our cognitive map will be of no use if everything moves around whenever we apply different criteria (Shum 1990). Assuming that virtual environments are more expensive to develop and to deliver, we must ask ourselves what we have gained.

There may also be educational objections to the spatialising of all relations between objects. We have to remember that space in such contexts is a metaphor for other things. Do we risk imposing simplistic structures on the learner's thought by adopting the single dominant metaphor of space? Nelson's remark about metaphor in the interface is relevant here:

In the metaphorical approach, the metaphor becomes the central concept - the principle, if you will, to which all other aspects of the design must adhere. The problem is [...] that slavish adherence to a metaphor prevents the emergence of things that are genuinely new.
Nelson, 1990 (original emphasis)

In constructing VRs to house certain kinds of knowledge, we risk tying ourselves to a single model where alternative representations may be better capable of promoting new thoughts. These and related ideas are discussed in greater depth in Gere 1996.

The 'Homeopathic Fallacy'

Another important issue for educational applications of VR is a difficulty which has been highlighted in relation to hypertext, namely the 'homeopathic fallacy' (McKendree et al 1995). The objection is to a claim often made: that the graphic visualisation of a relationship between a set of concepts is necessarily educationally productive. Hypertexts map the relation between information objects by one method, and a spatial virtual collection does so by another, but the objection is that in either case the solution is based on a false argument:

We intend that a learner should develop a conceptual network of the relations between items of knowledge
We 'depict' such relationships, whether using hypertext links or spatial organisation in a Virtual Environment: essentially the author aims to make the structure manifest on behalf of the learner
Now we encounter the problem: the structure is manifest in the system, and we wish it to be in the head of the learner, but 'we don't possess a magic funnel which allows knowledge to be poured into our heads' (ibid).

A related problem is that of the learner's ability to critically appraise a representation. When we see an N-dimensional information-structure mapped as a series of two-dimensional screens, we are not only able to conceptualise the relations between neighbouring items without those relations being pictured on the display, but also to remember that the sequence as presented is indeed a representation - one amongst many. Could the very persuasiveness of a spatial VR representation be a problem: that learners might cease to see how the relationships between items is contingent, susceptible to revision and questioning - the very sort of realisation which HE should foster in learners?

We must not think that the construction of virtual realities is a common-sense operation in which it is obvious what choices should be made, one in which design has no part. Perhaps there is a danger that we might lose heart in the face of the multiplicity of choices described in this report. But the point is that - to an even greater extent than in the physical world - we can build anything we like. The decisions we make must be informed by a knowledge of the wide range of possibilities open to us, and by an understanding of the processes of perception and cognition. These include not only what we build, but how we represent it and how the user engages with it - above all what educational processes it promotes and why.

Contents

Graphics Multimedia Virtual Environments Visualisation Contents