Title

Resources for a Christian worldview

Basden/OOSVE

 

This is an unfinished paper, but it is included in this collection because it addresses a very different topic than the other papers do. The ideas in this paper were presented at the CPTS Working Conference in 1999 but have yet to be published. So it is offered in its unfinished state in case the ideas are found useful. It used Dooyeweerdian ideas to try to understand the ontic status of things in virtual environments (virtual reality), such as characters and objects in a computer game. It employs Dooyeweerd's notion of enkapsis. Much of this can also be found in chapters IV and V of the author's book, 'Philosophical Frameworks for Understanding Information Systems'.

On the ontological status of virtual environments

Andrew Basden,
Centre for Virtual Environments, University of Salford, UK.

ABSTRACT

A perennial question in information technology research communities is the nature and ontological status of the various types of entities that present themselves, and the relationship between them. For instance, how do we account for the following, and in what ways can we say that they 'exist': hardware components, memory blocks, shapes on computer screen, programs, data structures, models, the knowledge that is encompassed in them, and, being discussed in more recent times, the people that use it? These questions are pertinent to virtual environments, where the not only is the range of different types of things more varied but it is closer to the users.

Dooyeweerd's radical view of entities, relationships and modal aspects seems to address these questions with an elegance and 'fit' not found in any other approach. This paper explores the application of his ideas, especially his concept of enkapsis, to questions of 'existence'.

INTRODUCTION

In what senses can it be said that a virtual environment exists? To what extent can it be said that any part of a virtual environment exists, such as the objects 'in' the environment? To begin to address such questions we must define which objects and things we are discussing, of which there are many.

For instance, suppose we are immersed in a virtual meeting (one application of the virtual environment concept and technology), in which we find:

  • ourselves and several others human-directed avatars, each made to look reasonably like the person they visually express,
  • several machine-directed synthetic agents (such as a virtual chairperson who keeps order in the discussion),
  • several objects such as a writing pads, white boards, etc, that have some function during the meeting (for instance, human participants can draw diagrams with their mice and the diagram appears on the white board),
  • several objects such as table, drinks glasses or vases of flowers that have no active function during the discussion,
  • all within some 'landscape' which in this case is a virtual meeting room.
These are all visual expressions of 'physical' objects; but in a virtual environment of this sort we might also have visual expressions of non-physical things, such as:
  • representations of some of the concepts of the ensuing discussion, perhaps as simple shapes floating around,
  • labels on each avatar and perhaps on other objects,
  • the content of the discussion, which will probably be conveyed aurally but might also be available as text along the bottom of the screen,
  • colour and texture of (parts of) each avatar or object showing something of their state (e.g. ready with question, ready with point of information, agitated, etc.),
  • the colour and texture of e.g. the table or walls adjusted to express the state of the discussion (such as whether it is in the phase of argumentation, consolidation, etc. (refce to IBIS?, 19).
  • animated sequences such as mouths moving in tune with what is being said.
  • In addition, a record of the entire discussion is stored in the computers and made available at the touch of a button in a window that is superimposed on the scene. So also is other relevant information.
Thus far we have a mainly visual scene, but we can widen it to include aural and tactile stimuli given to the human participants when various events occur, such as telling them there is a message for them. The actual discussion itself will probably be conveyed aurally. Each human participant will have one or more user interface control devices with which to interact with the environment, such as mice, datagloves, keyboards, microphones, etc.

We have a very rich environment of with which we, as participants, are involved. Not only do we have to detect, recognise and respond to each stimulus, but we have to control the user interface devices to take part in the discussion.

THE PROBLEM

After a period of learning, we become familiar with all the devices, sensory stimuli, etc. and take part in the meeting. At first we are aware of each object, but they soon become tacitly 'proximal' (Polanyi, 1967). However, in theoretical thought, we attend to each object or phenomenon distinctly, and the question raises itself: which of these objects can be said to 'exist'? That is, we are concerned with the ontological status of all the things we might come across when we have a virtual environment. If we analyse the situation we find there are many different types of 'existence', if indeed they can be called 'existence'.

The human participants 'exist'. But do their avatars also 'exist'? Knox (1974) has argued for the real 'existence' of appearances of objects; can we treat the avatars as appearances of the participants? If the avatars 'exist', in what way is their existence different from that of the human participants? (We might suggest that the human participants have physical form and the avatars do not, but in fact they do: the computer screen, so the matter is not as clear as first might be thought.)

If we are satisfied with the existence of the avatars, what about the synthetic agents? Does the virtual chairperson 'exist', and if so, in what way? Further, do the objects that are used in the discussion 'exist'. Do those not used in the discussion 'exist', such as vases of flowers and tables? Does the landscape 'exist' in the same way as these?

When we turn our attention to the second group of objects, we have yet another type of 'existence' - the concepts that are important in the discussion and their symbolic representation as simple shapes, the labels alongside the avatars or other objects, the colours or textures that show states of the avatars. Does the aural rendering of the content of the discussion 'exist', especially after it has been heard? What about the ankmated sequences; do they 'exist'? What about the whole text of the discussion?

In other applications of virtual environments and virtual reality technology, non-avatar objects are much more important than in virtual meetings, and present yet further problems. Take the case of a training aid for surgeons (e.g. Avis, 19), which is a virtual human body (or part thereof) that the trainee surgeon can (virtually) explore with (virtual) probes, make (virtual) incisions, etc. There is only one avatar, the trainee, and the major part of the environment is the complex of soft tissues such as blood vessels, alimentary canal, muscle, etc. These are not distinct objects as were the table, notebooks, vases, etc. For instance, an artery can become a junction, and there is no distinct boundary between them. So to what extent can we say that these 'exist'? Further, each of the organs have properties which are more important and more numerous than properties of meeting objects; in what way do these properties 'exist'? When an incision is made, what is its ontological status? To make matters worse, continuous material like the wall of an artery has properties of elasticity which vary along its length and circumference in a complex way; is the 'existence' of such continuously varying properties different from discrete properties?

If a window containing the whole text of the discussion, or some other information about the discussion, is brought up by one of the participants on their own screen, in what way can that be said to 'exist'? It does not appear on the other participants' screens, and (depending on exactly how the virtual meeting is implemented in software) it might be purely local to that participant, so that the central controlling software is not aware of it. ('Aware' should, of course, be placed in inverted commas there.) So is the existence of such a local phenomenon different from all the others, that are more truly part of the whole meeting?

Though we have posed questions concerning the existence of each of the objects or phenomena listed above, we have not come to the end of the existence questions for which we must find an answer. First, the avatars, objects, landscape, text, etc. that we see are all made up of shapes on the screen. Those that we hear are made up of aural 'shapes' known as phonemes. Animations are made up of sequences of shapes over time. Can these shapes be said to 'exist'? The shapes are made up of pixels on the screen; can these be said to 'exist'? Some might say that 'only' the pixels 'exist' and the perception of shapes and then the objects is merely part of the human process of interpreting these pixels. But a physicalist would go deeper and say that the pixels themselves do not exist; what exists is the physical blobs of luminescent materials which when activated emit light that our retinal receptors detect as pixels.

The existence questions continue in yet another direction, to the internal memory and workings of the computers that support the operation of the meeting. In most (all) circumstances, data structures that are internal to the computer are used to hold information about the visual (or aural etc.) shapes on the one hand and about the objects they express on the other. For instance, the boundary and colour of each visual shape must be held in such a data structure. Which shapes link together to form a single visual object that expresses an avatar must be held in data structures. (This is actually made more complex by the fact that we are dealing internally with three dimensions but the shapes seen on screen are a two-dimensional rendering of that; this distinction will be overlooked here.) The identity of the human participant that the avatar represents, together with information about whether it is their turn to speak, what they last said, etc. must be held in data structures. All the conceptual information shown, or that could possibly be shown, on screen must be held in data structures. Except for what the database community calls virtual information (a different usage of the word 'virtual' from that in 'virtual environments'), by which they mean information that does not currently 'exist' but could be calculated at will when needed. Statistical information, such as the number of utterances per human participant, is of this kind. The question we must face is: to what extent can it be said that these data structures 'exist'?

(Notice the use of the words 'express' and 'represent' above. In this paper we will use 'represent' to refer to the link between the real world thing (such as human participant) and its symbolic manifestation in the computer system (the avatar), and the word 'express' to refer to the link between the symbolic manifestation and its visual (or aural, etc.) appearance, in this case the shapes on screen that 'are' the avatar. In making this distinction, we are perhaps pre-empting some of our discussion about the ontological status of these things.)

Each of the data structures is composed of collections of 'bits' in the computer memory (which are normally aggregated into bytes of 8 bits each). In what sense can it be said that these bits and bytes 'exist'? That question is reminiscent of that about pixels above. In a similar way a physicalist would contend that neither the data structures nor the bits 'exist'; only the electronic gates that when activated can be interpreted as bits.

We notice that we have here two parallel sets of questions, one set concerning the user interface, about avatars and objects, the shapes in which they are expressed, pixels and material which provides the pixels when it is activated, and the other concerning the computer memory, about data structures, bits and electronic gates. In both we have come down to the level of manufactured physical hardware, and that leads to two final questions.

One is that the hardware is more than pixel-supporting blobs or memory gates, but must also include such things as the whole screen which the user sees, the mouse or dataglove the user wields for interaction, the box in the corner that is the computer that drives it all, and the network that links the computers. To what extent can each of these be said to 'exist'? It might be argued that, when we are considering the interaction of avatars or the interaction of data structures, then the hardware device becomes unimportant and so could be said not to 'exist'. But it still does 'exist' in some other way; what way? The question could be extended to the network between the computers: even if we are considering the dataglove being wielded by one human participant, we might be unaware of the network that links the computers, so that it does not 'exist' for us.

The other final question is: do the manufactured blobs of material and the manufactured electronic gates 'exist'? Or is it only that the materials from which they are made - silicon, copper, etc. - can be said to exist?

Beyond the final questions, however, we must still ask an overarching one, to question the assumption with which we started: To what extent can we say that the human participants 'exist'? We are not here talking about whether there are actual participants or not, as for instance whether they have all left their computers to visit the coffee room or whether they have suddenly died; while such suppositions are interesting and should be addressed at some stage, let us ignore them here, and suppose all the human participants are actively involved at their screens. What we are asking is: how does their type of existence impinge on and relate to the types of existence of all the other things we have discussed?

We can see that we have a considerable array of existence questions when faced with a virtual environment. Fig. 1 attempts to map out the various major classes of things we have applied the existence question to. I have added one further class: content, and linked it with a dotted line to Human participants.

  Human      ----------------- Avatars ------------- Visual (aural)
  participants               { Synth.agents          objects
     .                  ,----{ Mtg. Ojs.      Landscape |  (incl.
      .                /     { Non-Mtg. Objs    /       |  anims.)
       .              /          /   |         /        |
        .            /          /    |  Env'ment        |
         .          /    Properties  |     /            |
          .        /       /   \     |    /             |
           .      /  Concepts   \    |   /              |
            .    /  /Labels \    \   |  /               |
             .  /  / Text    \    \  | /                |
              ./  /           \    \ |/                 |
            Content            Symbolic                 |
                               representation           |
                                /   (Model)             |
                               /                    Shapes
                              /                    /  |
                             /                    /   |
                         Data           D.structs/    |
                      structures        of shapes     |
                            \         /               |
                             \       /                |
                              \     /                 |
                               \   /                  |
                               Bits                 Pixels
                                |                     |
                Computer,       |                     |   --- Screen,
                Disk,           |                     |  /     mouse etc.
                Network ----- Gates etc.            Blobs of
                                  \                lum. material etc.
                                   \               /
                                    \             /
                                     \           /
                                      \         /
                                   Silicon, copper, etc.

Fig. 1. Classes of things about which Existence Questions are asked

Specifically excluded from our discussion are questions about the following, some of which have already been noted:

  • Whether the human participants are actively present or not; it is assumed they are
  • The difference between the 'existence' of three dimensional shapes in the model and that of their two-dimensional rendering on screen
  • Whether the content of the discussion of the meeting refers to real or imaginary things in the real world, and if imaginary whether the imagined things are possible or not, and the ontological status of such things
  • The 'existence' of effects of having held the meeting
  • The 'existence' of control strategies for the meeting, such as how to let each participant know it is their turn to speak

However, for a full treatment of ontological status, these and other questions should also be addressed.

IMPORTANCE OF THE QUESTION

Why should we ask such questions about the ontological status of the various things we come across when involved with virtual environments?

TO BE DISCUSSED

# the question is interesting in itself, as philosophy # it can help clarification # OTHERS

ATTEMPTS TO ADDRESS THESE QUESTIONS

1. Hirst "Existence Assumptions" 2. Conventional Philosophy

==== to be written

MAKING SENSE OF ONTOLOGICAL STATUS

We wish to study ontological status of things we encounter in or around virtual environments. To do this, we must understanding what Being is. The Greeks presupposed that Being was the most fundamental property of all things, and that Being is one. That is, there is one mode of being, and within that mode there are many types of things.

[This following is partly for a non-Dooy readership.] Dooyeweerd takes a radically different starting point, in Meaning rather than Being, because he starts with a Creator. He says, "Meaning is the being of all that has been created" (NC, I, p.4). He expands on this: Meaning is the mode of being, the way(s) in which we can Be. In volume II of his New Critique he shows how Meaning has fifteen (or so) aspects, each of which is distinct and irreducible to the others, and yet closely interwoven. Therefore, if Meaning is the mode of being, and Meaning is in fifteen aspects, there there are fifteen ways of being. There are fifteen ways in which a thing can Be, not just one. (The number fifteen is Dooyeweerd's own proposal, but might prove to be a different number.)

Modes of Being

The possibility of different modes of Being seems to accord with our everyday experience. When I say "I am a husband", "I am a lecturer", "I am a Green", "I am a Christian", I am stating various modes of being, and all can be true without contradiction. Therefore the things we encounter in a virtual environment might also have different modes of Being, and can be different things at the same time.

The possibility of different modes of Being also gives both richness and freedom to our discussion of ontological status. It allows us to posit kinds of existence that are radically different from each other without being under pressure to reduce them to a single mode. We will take Dooyeweerd's proposal, namely of fifteen ways of Being, at face value and allow ourselves the possibility of different modes for the purposes of our discussion.

If there are different modes of Being, then how can this be so? Will we not just end up in a fragmented account that provides no help in understanding what is going on? Dooyeweerd proposed that not only are the aspects irreducible to each other, but that they are closely intertwined, and also that things integrate within themselves their various modes of Being. So this shifts the emphasis of discussion away from existence as such to the relationships between the different modes of Being. What is the relationship, for instance, between me as husband and me as lecturer?

Integrated Modes of Being

Dooyeweerd uses the term 'individuality structure' for a thing which is an integrated whole that might have various modes of Being. When it has several modes of Being, it is several 'entities' that are distinct yet intertwined. What conventional thinking labels an entity is in reality an individuality structure. Dooyeweerd's entities are qualified by the modal aspects which are modes or ways of being.

The possibility of there being several distinct entities within an individuality structure forced Dooyeweerd to account for the integrality of the individuality structure, and to do this he proposed the concept of enkapsis. The various modally qualified entities are enkaptically related; the first example he explored in his New Critique is Praxiteles' marble statue; it has a physical mode of being as a block of marble, and an aesthetic mode of being as a statue. As he developed the concept of enkapsis, it appeared that there were several different types:

  • Foundational enkapsis, as in a statue and the marble of which it is made, or in a cell and its physico-chemically qualified microstructures
  • Symbiotic enkapsis, as in clover and nitrogen-fixing bacteria
  • Subject-object enkapsis, as in a snail and its shell
  • Correlative enkapsis, as in a forest and its insects
  • Direct territorial enkapsis, as in a state and its orchestras
  • Indirect territorial enkapsis, as in a state and its tax-payers or its schools

Relationships

A sound concept of relationships lies at the root of any attempt to discuss different types of 'existence'. But conventional philosophy has proved largely inadequate to this task. Occam suggested that relationships are secondary to entities, and therefore should be excluded from any basic ontology, and this idea seems to have pervaded much thought down to this day, and has prevented any serious attempt to address relationships. Some, in reaction against this, have emphasized the importance of relationships (==== who has done so?), but have merely stated that relationships are contingent; the human interpreter can form any relationships they wish and no type of relationship has any ontological precedence over any other type. Relationships, to this class, are arbitrary in type - whatever we wish to make them. However, neither of these approaches suffices for our task. We have a variety of relationships here and though there can be some argument as to whether this is all we need to consider, we have something here that is neither secondary nor contingent. Many of the relationships are required by the situation, and we must give an account of them.

In keeping with his contention that Meaning is the starting point, Dooyeweerd proposed a strong and rich theory of relationships that might provide the account we need. It stems from his giving central position to Meaning as the To him relationships are primary and some relationships are essential rather than being contingent (for instance, that between a snail and its shell), and he was interested in delineating what these types of essential relationships are. There are several basic types:

  • Subject-object relationships
  • Part-whole relationshps
  • Enkaptic relationships

As we have seen, enkaptic relationships are perhaps Dooyeweerd's greatest contribution to debate about relationships, and their significance and the need for them arises from his view of entities and individuality structures.

DOOYEWEERDIAN EXPLORATION OF VIRTUAL ENVIRONMENT

His approach to relationships can help make sense of the plethora of types of existence we have discovered in virtual environment, more successfully and more naturally than other approaches can. We will therefore discuss the relationships shown in Fig. 1, and see how all can be accounted for.

Materials - Components (Gates and Blobs)

The relationship between materials and the components manufactured out of them is like that between the marble and the statue: foundational enkapsis. The physical materials are fashioned into components for some purpose.

Components - Hardware Devices

By devices we include the larger hardware devices that are of significance to the user of the virtual environment, including, on the left in Fig. 1, the computer, network, etc. and on the right the screen, dataglove, mouse, keyboard, microphone and sound system, etc. These are aggregations of individual components, and so we have a part-whole relationship, rather than enkapsis.

Gates - Bits, Blobs - Pixels

This would seem to be a type of foundational enkapsis. The individuality structure that is the statue is composed of two entities, and to talk of it as a statue is to talk aesthetically whereas to talk of it as a piece of marble is to talk physically. In the same way, to call the pixel a pixel is to talk sensitively (because we are emphasizing its being a meaningful stimulus of the human eye) whereas to call the pixel an activated blob is to talk physically. Therefore there are two entities in this individuality structure: a sensitive and a physical one. Note that pixels have colour, a concept relevant to the sensitive aspect, whereas blobs have only activation to a given wavelength. (In fact each blob is composed of three blobs one for each of three wavelengths commonly labelled red, green and blue.)

The same argument can be made for the relationship between activated gates and bits, except that bits are analytically qualified because the important thing about them is their intentional discreteness, their meaningful distinction from other things. Bits are the basic piece of 'information' (in a Shannonian sense) and can exist in two states: on and off, commonly labelled by the digits 0 and 1. (? Maybe they are quantitatively qualified, namely being discrete rather than distinct? DISCUSS)

Note however, the importance of activation in both cases. When the computer is switched off, or better when it is partly dismantled, the gate and the blob are both purely physical and they are not at that time also memory bits or pixels.

TO BE DISCUSSED: HOW DOES THIS IDEA OF ACTIVATION FIT INTO ENKAPSIS?

Pixels - Shapes

Shapes on screen are aggregations of pixels. For instance, a line is a set of pixels with certain positional relationships between them, and a certain colour pattern. Simple shapes tend to have pixels of all the same colour while complex shapes have variations in colour across them. However, the concepts with which we talk about shapes are from the same aspect as those we can apply to pixels: colour, position, etc., and all come from the sensitive aspect. Therefore the relationship between pixels and shapes is not enkaptic but part-whole.

DISCUSS: IS MY REASONING CORRECT HERE?

Bits - Data Structures

(Much of the explanation here is for Summer School; it would not be needed for a paper in Virtual Environments.)

The relationship between bits and data structures is similar to that between pixels and shapes but a little more complicated. Partly this is because different meanings are attached to the words 'data structure'. The meaning we will assume here is made clear in the following discussion, and indeed it is the use of Dooyeweerdian concepts that helps to crystallize its meaning.

It turns out to be enkaptic because there is another entity involved here that has not been named so far. It is the block of memory. It was said above that bits are commonly aggregated into bytes. In fact, there are several aggregations, into bytes (8 bits), nybbles (4 bits), words (16 bits), longwords (32 bits) and memory blocks (any number of bits but usually more than 32 and usually a multiple of 8 or 32 in size). (Let us call them all blocks (e.g. a byte happens to be a block 8 bits in size).) In computer science and programming, blocks are 'allocated' memory 'space', either by calling something like the malloc() function of the C language or from specially pre-allocated larger blocks called the stack or heap.

There is therefore a part-whole relationship between bits and the block to which they belong.

Normally a block is allocated for the express purpose of holding a data structure. The use of the concept 'data structure' implies a certain meaning placed upon the bits. Suppose we have a very simple data structure, a single letter of the alphabet. By convention, each letter of the alphabet is implemented in a block one byte in size, and each different letter (or digit, punctuation, or various things) is implemented by a different bit pattern. e.g. The pattern off-on-on-off-off-off-off-on implements the letter 'a' when using ASCII code. It is the code employed that makes the link from bit pattern to data held in a data structure. Using another code (say binary arithmetic) the same bit pattern would be interpreted as the number 97. So, while we talk in terms of patterns at the bit level, we talk in terms of symbolic values when we start to talk about data structures, and the relationship between them is that some meaning has been applied via a code.

Therefore the relationship between a memory block and a data structure is one of foundational enkapsis. The block is analytically qualified while the data structure is symbolically qualified; two entities within one individuality structure.

Note, however that most data structures are complex, being collections of various simple structures (called variables or fields) like letters, numbers, truth values and other things. The size of the block that holds such a data structure is often the sum of the sizes of blocks holding its parts. The relationship between a simple variable and its structure is of course part-whole, so we have two parallel part-whole relationships here [DISCUSS; IS THIS SIGNIFICANT?]. Further, a data structure can be part of a larger one, and the mechanism by which it is part can be of two types. Either the block can be within the block of the larger one, or it can be 'pointed to' by the larger one, and one then has two separately allocated blocks that are linked by a 'pointer' (which is usually a longword whose bit pattern is interpreted as a memory location).

Note that a data structure need not be fixed in its composition; the data structure for the whole discussion of the virtual meeting will comprise those for each of the individual utterances, of which there is no fixed number. In fact the data structure will grow over time. As it does, more and more memory is allocated to it.

Thus we have defined 'data structure' at what Newell (1982) calls the symbol level, at least as far as values are concerned, and have shown that there is another concept between it and bits, namely the memory block, which is at what Newell calls the bit level. Some people tend to define 'data structure' as something nearer to the memory block, not least because most data structures contain pointers to others and these pointers are bit level phenomena (an example is given below). If we were to define 'data structure' as at the bit level then we would find a part-whole relationship from bit to data structure and an enkaptic relationship from data structure to symbolic representation. However, as we shall see below, there is more to this picture.

Newell's (1982) paper was seminal in the artificial intelligence community. His main arguments need not concern us here, but he proposed a number of levels that are irreducible to each other, among them being the bit level, symbol level and knowledge level. He made the claim, astonishing at the time, that the knowledge level (which we will return to below) is not just a way of looking at things, it is "a fact of nature". That is, he was claiming ontological status for the knowledge level, and indeed all levels, which thus places his ideas close to those of Dooyeweerd. It seems that Newell's levels, which we can detect in the vertical progression in Fig. 1, have a lot of foundational enkapsis among them.

Data structures - Shapes

The relationship between data structures that hold information about the shapes on screen and the shapes actually on screen is interesting. Suppose we have an N-sided polygonal area of colour 'lemon-yellow'. A typical data structure for this would comprise:

  • An indication that this shape is a polygon (rather than for instance a line or a circle)
  • N triples of (x,y,z) coordinates, one for each vertex of the polygon
  • An indication that the colour is simple (rather than being shaded, for instance)
  • Three numbers showing how much of each of red, green and blue the colour consists of (more green than red and a little blue, in our case)
Those are the main parts of the data structure, but in most cases others are required for administrative purpose within the program:
  • A pointer to the next shape data structure, which is required by the part of the software (or hardware) that actually draws the shapes on screen,
  • A pointer to another data structure which in our case would be the data structure for the avatar or other object that this shape expresses; this is necessary for when the user clicks with a mouse over this shape: the software must be able to find out which things the shape expresses.

What is the relationship between the data structure and the shape visible on screen? It might be foundational enkapsis, but it 'feels' different. Is it a non-biological kind of symbiotic enkapsis? [DISCUSS]

Data Structure - Symbolic representation (Model)

Symbolic representation is very similar to what we have defined as data structure, comprising numbers, text, truth values, etc. among other things. Therefore, so far, we can think of them as almost synonymous. If we had defined 'data structure', as some do, as more like memory block, then there would have been a foundational enkaptic relationship between data structure and symbolic representation. So are they really synonymous, when we define data structure at the symbol level?

The answer is that symbolic representation is richer than our definition of data structure. Data structures see the world in terms of variables (fields) and values held in those fields. Other names for the fields include 'attribute' or 'slot', but the idea is clear: these things are containers of values (which might be quantitative or qualitative in nature). Now, symbolic representations normally have a richer representation formalism, namely entities and relationships in addition. The object-oriented and frames formalisms also add procedures.

Basden (1993) called these aspects of knowledge, and discerned four types that are irreducible to each other:

  • items (entities) and relationships
  • quantitative and qualitative values
  • spatial knowledge
  • change: events and processes (procedures)

To the extent that a representation formalism omits or ignores any of these, to that extent is it limited in its 'appropriateness' as a knowledge representation mechanism, and to that extent will its user find difficulty. It is certainly possible to implement one aspect by means of another, but various difficulties arise.

In this light, data structures, which are mainly limited to the second aspect above, are a limited form of symbolic representation. However, there is one natural way to add items to a data structure, namely to treat each data structure as implementing an item, and its parts as implementing the item's attributes that hold their values. This has been recognised by the popularity of object-oriented languages with which software is now written, which adds items and (in a limited way) relationships. In fact, the picture shown in Fig. 1 is perhaps a little old-fashioned, because much virtual environment software today is written in an object-oriented language.

Therefore a particular symbolic representation will be implemented by means of one or more data structures in ways that are more or less appropriate. For instance, if the data structure mechanism does not directly support spatial information, then it must be implemented (possibly by means similar to that described in the previous section, yet far more complex).

Therefore we can say that the relationship between data structure and symbolic representation is not fixed. With highly appropriate data structuring mechanisms, covering all four aspects above, then they are largely synonymous. But where implementation is required, then we have what appears to be a foundational enkaptic relationship.

There is a foundational enkaptic relationship between symbolic representation and memory blocks of which there are at least four types, depending on the number of aspects of knowledge. Data structure could, therefore, be a half way house between memory blocks and symbolic representation in which some of these enkaptic relationships lie between it and blocks and others lie between it and symbolic representation.

DISCUSS THAT.

(Another possible difference between symbolic representation and data structure is that the symbolic representation need not be by means of data structures in the sense we have defined above, that is something based on memory blocks. Instead, it would be written or spoken text, drawn diagrams or hand signals. However, the type of representation shown in Fig. 1, that is for a virtual environment, is based on computer memory, and hence this difference is not significant here.)

On a note of terminology, we will use the term 'model' to refer to the whole symbolic representation.

Shape - Visual Object

Shapes are aggregated into visual objects. There is therefore a part-whole relationship between them. This is also true in the case of aural and tactile sensory objects, by a similar argument as we have used for visual objects.

However, the case of animation requires a little discussion. Here we have an aggregation in time rather than in structure, so again we have a part-whole relationship. That is, each frame of the animation is part of the whole. When animation is used as a symbol for some meaning, it plays the same part as any other sensory object; see below. But there is a difference that might affect at least the technology and how it is employed in real situations: frame animation is only an approximation to true animation imposed upon us by the discrete nature of computer technology. True animation is continuous, in both space and time.

Symbolic representation - Avatars, Synthetic agents, etc.

In our discussion of symbolic representation above, it becomes clear that avatars, synthetic agents, meeting objects and non-meeting objects are just items within a model (a whole symbolic representation). They are mainly of the items and relationships aspect of Basden's four aspects, though rich in structure and involving the other aspects. Their attributes contain values, the items have behaviour, namely events and processes, and they have spatial properties. The avatars and the synthetic agents can be considered (symbolic representations of) functioning subjects; the meeting and non-meeting objects are not.

As functioning subjects, avatars are closely tied in with the human participants which they represent, and this relationship is discussed below. However this cannot be said of synthetic agents like the virtual chairperson.

What is the synthetic agents' status as functioning subjects? In which aspects do they function as subjects? Strict Dooyeweerdian thought would say they function only physically as subjects since they have no biotic reality. The argument would be that they do not really 'exist' as symbolic things [WOULD IT? ====] nor analytical bits, but only as activity of physical gates in the electronics of the computer (and, in parallel, as activity of the luminous blobs on the screen). However, such a treatment would seem to reduce their real meaning.

DISCUSS: WHAT IS THE STATUS OF SYNTHETIC AGENTS AS FUNCTIONING SUBJECTS

Symbolic representation - Concepts, labels, etc

The same can be said for concepts, labels, utterances and other entities whose real-world referents are conceptual rather than physical. The main difference is that they do not (normally) have spatial properties, and they are not normally active, that is functioning, subjects. In most other respects they are to be treated the same.

Symbolic representation - Properties

In the same way it becomes clear that properties part of a whole symbolic representationl; they are essentially attributes that hold values. Therefore again a part-whole relationship exists.

However, properties are often richer than the simple attributes of Basden's aspects of knowledge might suggest. While each distinct object has a number of properties, the example was given above of the continuously varying properties of soft tissue like the wall of an artery. This requires good support for spatial knowledge in the symbolic representation formalism offered in the virtual environment. (It is no coincidence that Dooyeweerd's kernel of the spatial aspect is continuous extension; it is this very thing that poses problems for knowledge representation in such cases.)

Symbolic Representation - Environment

What of the environment in which the virtual meeting takes place? In conventional systems science this is considered just another system, like all systems (complex entities that have a part-whole relationship with each other). But Dooyeweerd maintains that things like society, environment cannot be treated as individuality structures. [HAVE I GOT THAT RIGHT?] However, the environment, in our case the virtual meeting room, can have attributes that are meaningful, such as the colour of the walls and any background noise that gives 'atmosphere'. Therefore, the environment can be represented as part of the model.

The relationship between individual pieces of symbolic representation and the virtual meeting room would seem to be an example of correlative enkapsis. In this kind, the environment and its inhabitants are separate individuality structures yet each depends on the other for its full meaning. In the example given above the forest insects cannot exist (for long) without a forest, and the forest would deteriorate as a forest without its insects. It is even clearer in the relationship between trees and forest. Likewise, the participants of the virtual meeting need a virtual room in which to function, and the virtual room is meaningless as a meeting room unless it has (or is waiting for) participants. Hence this appears to be correlative enkapsis.

Representation Objects - Visual object

In general, each object or property in the representation, including the environment, has a visual (or other sensory) expression. Typically in a virtual meeting, the main functioning objects and the concepts are shown as distinct visual objects. (However, in the case of the surgical trainer the sensory stimuli are not sensory objects so much as sensory fields, such as the feel of the elasticity as the virtual probe moves along the vessel.)

What type of relationship exists between sensory objects or phenomena and the representational objects they express? As with the relationship between data structures and shapes, it would seem to be a kind of symbiotic enkapsis that is not biological in nature.

DISCUSS: IS THIS CORRECT? ARE THERE DIFFERENT TYPES OF SYMBIOTIC ENKAPSIS THAT DIFFER ACCORDING TO DIFFERENT ASPECTS?

Content - Concepts, Objects, etc.

Especially the conceptual objects in the model have external referents that are their semantic content. So, for instance, if there is a representational item labelled 'poverty' (shown as a brown cube in the middle of the table), then it refers to poverty itself (or, at least what the participants conceive of as poverty). In the same way, a meeting object like a whiteboard on which appears diagrams that are drawn by human participants will represent, not so much a real whiteboard, but rather the diagram that is currently being discussed. The utterances represent their own content. And so on.

However non-meeting objects and some aspects of the environment (landscape, meeting room) do not represent anything in the real world. Such things are mainly included for purposes of decoration or aesthetics. [HOW ARE WE TO SEE THESE? DO THEY 'EXIST'?]

The relationship between a symbolic representation (with its visual expression) and that which it represents would seem to be foundational enkapsis, but is it? Dooyeweerd says of a book or music score (NC III p.110, his emphasis) "They can only signify the aesthetic structure of a work of art [in our case, of the thing being represented] in an objective way and cannot actualize it." (Note, Dooyeweerd's use of the term 'signify' is equivalent to our term 'represent'; his use of 'represent' in NC III p.151 is subtly different.) Dooyeweerd's discussion of books on p.151-153 on NC III seems relevant to much of our treatment of virtual environments. Dooyeweerd discusses the content of books, but I am not yet clear what he says of it.

[DISCUSS: WHAT DOOYEWEERD SAYS COMPARED WITH WHAT WE ARE SAYING. I DON'T GET MY MIND ROUND IT ALL]

Avatars - Human participants

The avatar could be said to represent its human participant, just like any symbol represents something. To some extent this is true, and if I, as a participant in a virtual meeting, see an avatar, I can think "That avatar represents Joe." What this representational relationship is, is discussed in the next section. However, that would only occur if I became somewhat distal from the meeting; when I am fully involved then I think of Joe's avatar as Joe himself. So perhaps there is something richer than mere representation going on here.

The word 'avatar' is from Hindu mythology, where it means the body that a god takes on when it comes to earth. In virtual environments, the avatar is the virtual body or form a human takes on when s/he becomes involved in the virtual environment. Therefore this could be somewhat more than mere representation. Typically, in a virtual meeting the avatars are not precise visual likenesses of their human participant, but are designed to be easily recognisable as such. Various methods exist of ensuring this, such as showing them in different colours, perhaps with a side-bar which matches colours to snapshots of actual participants, or placing the name of the participant as a label above the avatar's head, or wrapping the picture of the face of each participant round a parameterized avatar head so that the avatar's head looks like that of the participant.

Typically the avatars are seated round a table, and the view each human participant sees is of the same virtual group but from a different perspective. In Microsoft's recent version, each human participant has a view from immediately behind their own avatar, so they see the top of their 'own' head, with all the other avatars facing it. When one avatar speaks, all the avatar heads point towards that.

What is the relationship between the human participants and their avatars? At first sight it might be a subject-object relationship, in that the human participant takes action which affects the avatar among other things. But this is different from the true subject-object relationship that exists between the human participant and for instance all the other avatars when they speak. It seems as though the relatioship between a human participant and avatar is more like that between a snail and its shell. The shell is enkaptically bound to the snail by subject-object enkapsis. In the same way the avatar is enkaptically bound to the human participant by subject-object enkapsis.

DISCUSS: IS THIS CORRECT?

Content - Human participants

Fig. 1 shows a dotted line between human participants and content. This is to signify that they are involved with the content and that it means something to them.

[I'm not sure how to treat this; DISCUSS.]

CONCLUSION

We have found that many different kinds of things (objects, phenomena, etc.) are encountered in virtual environments, and that Dooyeweerdian concepts can account for the relationships between them. Given Dooyeweerd's proposal that Being is not fundamental but that there are several modes of Being that are dictated by a diversity of Meaning, then understanding these relationships helps us answer the question as to which of these things 'exist' and in what way they can be said to exist. In particular, various individuality structures that are to be found in a virtual environment (such as the avatar) take on several different existences in different modalities, and these existences are integrated within the individuality structure by relationships of foundational enkapsis. Other types of enkapsis occur between the symbolic representations and their visual (or other sensory) expressions on screen, and between the human participant in a virtual environment and his/her symbolic representation as an avatar.

This exploration has necessarily been brief and limited, and many issues have been omitted or skated over. In particular, we have not considered the ontological status of the user's activity at the user interface. Other issues which have been omitted from discussion have been noted earlier.

However, applying Dooyeweerd's ideas to virtual environments has shown sufficient promise to be taken further. A number of points have been raised for further discussion.

First, it seems that the enkaptic relationship between memory blocks and symbolic representation displays a multi-threaded nature, in which there are at least four 'aspects of knowledge', for each of which there must be an enkaptic relationship - as though there are four enkaptic relationships. This gives rise to the possibility of an intermediate concept, the data structure, in which some but not all of the enkaptic relationships exist between it and lower level memory blocks. Those that do not exist must then occur between it and the higher level symbolic representation. Whether this idea of multi-threaded foundational enkapsis has validity must be tested by further philosophical research.

The second issue raised is that there seems to be a type of symbiotic enkapsis between things in computer memory and their sensory expression at the user interface. Normally symbiotic enkapsis is between biotically qualified things, but not in this case. Should the idea of symbiotic enkapsis be expanded into non-biotic areas to account for these types of relationships?

The third issue is that Dooyeweerd's concepts might help explain Newell's (1982) levels for computer systems, which range from hardware up the knowledge level, and which Newell claimed was "a fact of nature" rather than merely a way of seeing things (that is, an ontological rather than epistemological statement). Newell did not attempt to define the relationships between levels, but a combination of Dooyeweerd's concept of enkapsis and the part-whole relationships seems to account for his levels within two vertical sets of relationships, both computer memory and on screen. The pattern of alternate application of part-whole relationship to achieve aggregation followed by foundational enkapsis to move up to the next level is worth further exploration.

REFERENCES

==== to be supplied


Created: May 1999
Copyright (c) Andrew Basden. 2007

Last updated: 12 December 2007