Information Design Institute
We cognitive agents, humans and computers, currently interact with and process information in order to adapt meaningfully to the world, both directly and through our representations of the world, the latter built up in the coherent pictures we call common sense and science. Can that coherent basis of information lead to an autonomous Leibnizian processing in which information interactions take place outside of our current human structures altogether? Such a prospect would add a totally new dimension to cognitive ergonomics.
Keywords: information ontology, evolutionary realms, computing, disembodiment, emergence.
Computing is transforming itself into a digitised socio-informational virtual world in which we will interact with information, tasks and autonomous agents more than with our physical environment. Not only will our current interfacing with computers adapt to this transformation, but the very form of information (how we design information) is likely to change as well.
Current technology is positioning us on the doorstep of an information realm that portends of vastly different interactions and modes of being in the world. As agent technology continues to expand the scope of artificial intelligence and impact human-computer interaction, we need to reconsider the nature of information and ask to what extent autonomous information processing, independent of human agency, is possible and how.
1.1 The question of autonomy
A question naturally arising about the information realm is whether it can prosper outside of human cognition, in digital information artefacts for instance, and what that might mean for human identity. A future in which humans and autonomous agents each partake in cognitive processing, both jointly and independently, is the likely scenario currently envisaged.
In its strongest formulation, the question addressed in this paper is the possibility of disembodied information. Specifically, can information exist and information processing take place without an agent (even an artificial agent)? What would that be like? Indeed, if it proves at all possible, what will that be like? What we want to explore in this way is whether it makes any sense to think of information and information processing occurring at some point in the future in a fully abstract and immaterial manner, i.e. without a physical substrate. The mental and intangible character of much information gives an inkling that such a state of affairs might well be a possibility. Countering this, on the other hand, is the current realistic grounding of all mental processes in a physical medium (no cognition occurs without a brain or a computer). The question remains open and it is our purpose to explore it here.
The less drastic question, less pregnant philosophically as well however, concerns agent cognition - the issue of how we will interact with computer information agents that autonomously process information and interact with both humans and other information agents. The impact on the field of human-computer interaction is explored in Duchastel (2002). While this latter issue is of greater importance in the near future of cognitive ergonomics, it becomes very secondary in the larger context of full information autonomy, involving independence from any agency whatever.
1.2 Importance of addressing autonomy
The autonomy question opens up fundamental questioning of a philosophical nature, both in general terms (in ontology and epistemology), and specifically in information philosophy. The latter set of concerns deals with issues regarding the nature of information, knowledge, information processing and processors.
We are brought to consider ontological questions, particularly the interplay between thing and process, a far from straightforward issue. While of intrinsic theoretical interest, the concerns are also very practical in that much of cognitive ergonomics, embodied in computer interfaces, rests on human metaphors. Exploring the deep issues underlying information can lead to new metaphors to enhance cognitive ergonomics.
With autonomous information processing growing through agent technology, these theoretical considerations are likely to strongly impact information interaction, and hence the larger context of cognitive ergonomics.
Indeed, what I am addressing here is essentially the context of cognitive ergonomics, within the much larger picture of information interaction. We are not focusing on improving human interaction with the cognitive environment, but rather on what that environment will evolve into over the course of the near future. The tone is futuristic, and hence speculative, as well as philosophical in its consideration of the cosmological, evolutionary, context of the universe. We delve here much beyond human ergonomics, although only to better see the direction in which human ergonomics is likely to evolve.
And while focusing on interaction, this adventurous exploration necessarily brings us around once again to enquiring about the very nature of information, and hence the very substance of cognitive ergonomics.
2. Universal evolutionary context
From an evolutionary perspective, one can see the gradual emergence of different realms of interaction in the universe (Duchastel, 2001):
2.1 The evolution of realms
Realms are strata of being, modes of existence which enable (and cause) certain interactions. We shall end up examining pure information interactions in the informational realm, but let us see how it all came about and gradually evolved. The story to be told is fascinating and revealing, particularly the later stages.
The big bang at the origin of the universe was a release of energy which soon led to the creation of matter and much later on to the formation of the galaxies, planets, our own Earth and all its structure. That physical realm involved only physical interactions, ordered along the lines of what science is discovering as the laws of physics.
Circumstances were just right at one point in time on planet Earth for very simple life to emerge from those interactions, heralding a new realm that vastly complexified interactions and created, through biological evolution, a vast gamut of structures. Biological interactions (processes, laws) built on top of the physical ones. They did not change the latter (indeed, they are constrained by them), but the universe, in this small part of itself, grew in complexity.
Darwinian evolution continued to play out in such a way that cognition evolved naturally from simpler forms of biological interaction, eventually providing humans with our current mental abilities of memory, thought and consciousness. As with life (the transition from the physical to the biological), cognition (heralding in the mental realm) was a gradual affair, helping to explain its emergence from and its continuing basis in biology (Dennett, 1996). Cognition, i.e. representations and their processing, adds a whole new dimension to the world, while building on and respecting the laws that preceded it.
Cognitive interactions brought about culture and technology, again through a gradual evolution (Donald, 1991), that however fed upon itself and greatly accelerated the pace of change (what we typically call progress). Half a century ago, this led to the invention of the computer, an artificial symbolic processor that currently engages in memory and thought (but not in consciousness, yet).
The virtual realm (Levy, 1998) emerges out of the digitisation of the world, i.e. out of our current ongoing effort to represent all we see, know and imagine in digital form. The interactions with digital artefacts and with virtual worlds does not change the interactions in lower strata (the laws of physics, biology and psychology still apply), although massive change occurs in how we interact within our culturally constructed world (how we seek information, for instance, or how we relate to one another, or how we spend our time, and so on - van Dijk, 1999).
The virtual realm, while keeping a foot in previous realms (and constrained by their laws, at that level), creates parallel worlds in which the laws of the universe may be modified at will. I may thus fly over the land, become super-strong, and forego my identity. The imagination soars, while coherence takes a hit (in these parallel worlds, that is - back on Earth meanwhile, I still have to eat and sleep, and so on). Complexity has just increased, once again, in an astounding way (and the rate is accelerating). Where this will lead, socially and culturally, is an open question.
2.2 The informational realm
The last of the five realms envisaged, for now, is the informational realm. The informational realm is more than a speculative future; it is a philosophical exploration. Here is what is foreseen - and our task will be to see how plausible it can be. The scenario:
At some point, knowledge (a structured set of symbolic representations) starts interacting within itself, directing the process in the creation of new knowledge. Where this will go is a fascinating mystery.
As life emerged from and overlaid matter, and as cognition emerged from and overlaid life, and as the virtual emerged from the mental and broke out of the bounds of coherence, so too, the informational will emerge from the virtual and extricate itself from the constraints of the lower realms.
To be sure, computer agents are already actively transacting numerous interactions in the service of controlling processes such as financial transfers or manufacturing tasks. And they are increasingly serving as personal assistants to help users find the best deals in online shopping, and so on. But their autonomy remains limited and they are kept in close check.
That, however, will change with progress in technology (D’Inverno & Luck, 2001). As for the discovery of new information by computer agents, that is already a given in the field of machine-learning (Woolridge, 2002) and it is a trend that will also grow with time.
Now, the most contentious part of the scenario presented above is that abstract knowledge, which is structured information, starts interacting within itself. In a fully lawful world, where knowledge represents those relations between elements, a rational logic will naturally lead to new truthful conclusions, as initially imagined by the 17th century philosopher Leibniz. The logos of the world, what makes it coherent to us, underlies all interactions taking place in it. All of science is but the mapping by humans of those interactions and the inferential extension of new mappings that are harmonious with the established ones. That these can be made by non-human agents is readily acceptable. That they can be made by elements of information that are dispersed in different computer memory banks remains hypothetical, but highly plausible. To see how, we must first look deeper into the nature of information, in section 3. Likewise for the issue of information processing extricating itself from the constraints of the lower realms.
2.3 Beyond the human framework
The evolutionary transformation portrayed above is not anthropic - it does not centre the world on man. Humans play a crucial role in the mental realm, but then are displaced as evolution proceeds. One must adopt a non-anthropic perspective (difficult as that may be) to see the possibility of this scenario. In particular, one must drop the human normative stance (our set of human values) to go beyond the usual 'we won't let that happen' reaction. [Being unwilling to do so puts an end to this exploration, needless to say].
So if we look at this evolution not from our human perspective, but from that of the universe, what is happening? The universe starts off with a bang, develops elements (gradually all those in the Table of Elements), develops life and evolves cognition in some of the animals, creates technology that then allows knowledge to abstract itself out of the lower realms… on into the imponderable.
This is an agent metaphor, even if an impersonal one: the universe is something doing something in some structured way. Now for a life metaphor: the universe is an organic whole evolving in response to conditions within it. Consider even a physical metaphor: the universe is a grand process where one thing causes the next in an unending, unguided chain of determinism. Metaphors are tools of understanding, but we must beware of their appeal. It is always hard to remember the Kantian lesson in epistemology: however we know and describe the world, it is always through our human cognitive apparatus.
Despite this danger, metaphors are essential tools of comprehension and direct inspiration for design.
3. Nature of information
One aspect of the evolutionary framework that needs to be underlined as we seek to grasp the nature of information is the growth in complexity. This will lead us to consider the intangible nature of information, alongside other intangibles such as energy, life and consciousness, each defining a prior realm. We also need to re-examine the classic communication concept of information within a behavioural-functional context and consider the external storage function as leading to greater processing autonomy.
3.1 Complexification threading itself through the realms
Cutting across evolution of the universe is one growth feature that stands out: complexity. Starting with one undifferentiated singularity at the big bang, the universe is in a gradual process of local complexification. Mainzer (1997) provides the context for this phenomenon and numerous examples of it across the sciences.
The transformations that took place can be considered to all fit within a threefold context:
Complexification continues to grow as the various stages of evolution engage, indeed as the various realms of being come into being. Life for instance is a jump in complexity, as is consciousness, leading not only to a more complex universe, but also to what we consider to be different forms of existence. Our very human way of categorising being (alive vs. inert, conscious vs. reactive, etc.) is based on this complexity dimension.
An aspect of complexification, or perhaps better thought of as a requirement for complexification, is localisation. Complexity does not evolve uniformly in the universe. The very formation of the galaxies meant a process of assemblage, of merging together in a locality the elements that form them, and leaving relatively empty space behind. Localisation is thus a process of de-uniformization, of differentiation, of complexification. All these seemingly synonymous words point to an underlying dimension of being that cuts across the categories we have developed for talking about it.
Localisation involves a merging of things in the space-time-scale dimensions and hence gets into the zone of dangerous thinking involving Kantian subjectivism. Be that as it may, unfolding complexification and its process of localisation actually create the stuff of the realms, i.e. they embody: they lead to the formation of more and more complex bodies which capture and carry out processes.
In sum, the universe evolves towards greater complexity, each realm adding a new form of complexity to the current status of the universe at the time. Despite complexity-as-localisation being within a dangerous thinking zone, it can be considered to be the main thread through the realms, and is thus likely to inform our understanding of information interaction. Indeed, bodies emerging from the occurrence of complexification eventually create, sense, and manipulate information, becoming in effect information processors.
3.2 Communication and storage
Each realm has its intangibles: energy, life, consciousness are all philosophical terms in that they are difficult to pinpoint, even though they are easy to grasp intuitively. Information fits right into this pattern, as does the virtual likewise (Levy, 1998).
Information is a signal of a change taking place, for instance the distant rumble of thunder announcing the oncoming storm. Simple, direct, natural. At the other end of the spectrum, complex directions on how to assemble a kit car model are also a signal to guide action. Intricate, artifactual, directive, symbolic, knowledge-based, this information remains a signal.
The classic communication concept of information (sender-message-receiver-effect) is a useful perspective to bring out the relational nature of information. While both sender and effect often seem secondary (nature is sending the thunder sound and the effect is very relative), message and receiver are crucial. The message is information and the receiver (the perceiver) is what makes it important. If no sentient being hears the thunder (as was the case, for instance, before life on earth), then there is no information, just a disturbance dissipating as a wave pattern through the air. Crucially then, information requires some system for its interpretation, the latter being a defining characteristic of information. Even DNA, one of the classic prototypes of stored information, requires a biological interpreter [a forming cell] to make any sense of it.
An important distinction to make about information is the active-potential one. The thunder is very actual, live information so to speak, whereas the kit directions constitute potential information, captured and stored away in written form until needed. Knowledge, both personal and cultural, is of this potential kind of information.
The evolutionary perspective will help us understand the import of this. Information grows out of the action-reaction process in earlier realms. Physical and biological processes of a cause-effect nature (that being our way of relating them) take place in the world. Living organisms mediate those processes to their adaptive advantage, this being a characteristic of the life force they partake of. In other words, they position themselves advantageously in relation to the environment they are in. They may do so in as simple a way as orienting themselves towards a light source or in as complex a way as fighting their way up the social ladder of the pack or clan they belong to.
Cognition was a big step in this modulation power (Donald, 1991). Simple learning was definitely advantageous and as it grew in cognitive complexity, became ever more so. Representation and memory evolved in support of this line of development. Abstraction came about and knowledge took shape, initially only internally and thus simply, and then externally, in the culture, tremendously expanding in complexity of both form and process. Knowledge is in fact stored information, potential information that can be used at some later time.
We are at the point now where much of our mediation of the world, our interaction with it, occurs through information and knowledge stored and processed by computer (our invention for performing calculations, but which evolved into a machine for the easy storage and processing of symbolic information). Its predecessor, the book, provided storage but no autonomous processing. The computer, by merging data and program (content and process), enables information autonomy to take hold, a potentially crucial transformation for our issue of autonomous information.
Let us turn once again to the evolutionary perspective. Information was at one time merely active information (of the thunder kind) that informed action directly, in an ephemeral way - that is, it came and it went, part of an ongoing process. Cognition brought about mediation through the introduction of storage (memory and then culture) and processing (thought) that developed more and more abstractly and hence powerfully. As we have externalised information storage, so too more recently have we done so with processing. The upshot is that both information itself and information processing have become autonomous (to an extent) - becoming one step more removed from their previous embodiments.
3.3 Embodiment and change
Embodiment is the central concept we need to explore here. We return to the space-time-scale perspectives we encountered in discussing complexification and we immerse ourselves deeply into ontology (the question of forms of being).
Process and thing are the two principal perspectives we have for describing the world (the third, scale, is less relevant at the moment). These perspectives relate directly to our time-space perceptual apparatus in a true Kantian sense (namely that we perceive the world through these cognitive filter-enablers). Processes are events that flow through time and that transform things in space. Things are structures constituted in space that last over time, keeping a certain stability for a while. These are merely two ways of seeing the same phenomenon of change. We can look at change from a process perspective (the transformation of things - things changing in their spatial configuration) or from a thing perspective (the stability of things - their unchanging character between transformations). Both being relative (according to scale) on the time and space dimensions.
Embodiment is a different term for that thing/spatial perspective. A process results in a state of affairs getting fixed in a physical structure, in effect in a 'body' (in the large sense of the term, not necessarily biological).
Consider examples in the different realms. A volcano is a transforming process that fixes the surface of the world in mountains (in conjunction with other forces, such as plate tectonics and weather, and having other effects as well, such as forming lava fields, etc.). Digesting food is a transforming process that fixes the energy of ingested nutrients into the blood stream of the organism. Learning geography is a process that fixes information about the layout of the world into the brain of the person learning it.
The structures thus created as a process gets fixed into a state are physical structures. But there are other structures as well that result from processes. In biology, the structure is a functional one, one that embodies the role that the element plays within its larger context. In cognition, the structure is an abstraction, a relationship between cognitive elements that synthesises a new element representing that relationship. We can thus start thinking of embodiment within different structures, even though a physical substrate is always present.
Though consciousness is a fluid, philosophical concept, it is squarely embodied in the brain (despite questions about the nature of consciousness, naturally arising out of our current ignorance of the mapping between mental events and neurology). Indeed, there is no consciousness we reliably know of that is not embodied in a live brain. As for knowledge and thought, they are physically grounded in neuronal or electronic patterns - the quickness with which these change does not modify this fact.
The emergence of information is a consequence of complexification in the support of life. Processes in the physical world (what we know of as the laws of physics) were embodied in the shape of that world. Life mitigated those interactions by interposing adaptive behaviour between them (this being a biological process of complexification). Information is the stimulus for behaviour. Seen from the other side, the living organism reacts on the basis of information in the state of the environment. It responds to changes in the world. This is not unlike, even if at a different level, the billiard ball reacting to changes in its environment, such as the sudden impact from another billiard ball. In fact, information can be seen as leading to the emergence of all behaviour.
But consider the mediation that is occurring in the case of the organism. Its behaviour, even though the organism is grounded physically, distances itself from the physics of the environment, operating at a higher level, so to speak. It is as if, even though the physical constraints still impinge, the nature of the process (indeed, what is important about it) lies beyond those constraints. Organisms operate in a different plane of existence, defined as a different realm. Indeed, the basic difficulty of defining the realm-jumping concepts like life (those concepts that initiate a new realm) lies in our trying to describe them in the concepts of the prior plane, even though they essentially operate beyond that plane.
Proceeding onwards in the evolution, life began responding to ever more complex patterns of the environment (i.e. to more complex information) and soon began internalising these patterns in cognitions that could be processed and stored (Dennett, 1996). Once again, there is mitigation of the regular behaviour, with mediation this time being cognitive. As before, everything remains grounded, but the process starts operating in a new plane, what we call the mental. As with life, squabbles of definition result from our attempts at reductive analysis.
Now what we are seeing underway here is a process of disembodiment. While each new activity (life, cognition) remains grounded in the constraints of the previous realms in a true hierarchical fashion, the activity in its defining essence begins operating in a way that is beyond that grounding - in effect in a disembodied way in relation to the previous realm. That indeed is what disembodiment means - outside of a physical body or of a biological structure.
We are led in conclusion to question our very notion of processing autonomy, suggesting an enrichment of it as a mutation along a dimension (abstraction being a prime example, life another) rather than a mere localisation in space, hence only coupled to the physical realm.
4.1 Simple autonomy
Actually, information interaction of an autonomous fashion is not at all surprising given current trends in technology. As agent technology continues to expand its capabilities in symbolic computing and as machine learning continues to grow in sophistication, they will lead to the creation of new knowledge, including the creation of new processes. Some of these processes will be more advanced means of knowledge creation. This leads to a spiral of development that feeds upon itself and in which ever more sophisticated information interactions take place.
These interactions take place not mainly between humans, the current locus of knowledge development, but between computer agents to which humans will have delegated such tasks. In effect, then, these interactions will take place within the information realm, given that these agents are but information structures (computer programs). In the end, then, we witness the autonomous creation of information by information artefacts, initiated by humans and somewhat controlled by them, but nevertheless independent of them in intellectual exploration.
This is not dissimilar to what has occurred in other realms. Life emerges out of the physical realm and is continually nourished by it, but nevertheless follows its own pathways of development in growth at the individual and evolutionary levels. Likewise for consciousness, as it emerged from the biological realm.
4.2 Relocalised autonomy
As complexification increases, the locus of agency is displaced upwards, that is, in the more complex structure rather than remaining in the simpler elements before the structure emerged. Thus, interactions initially took place between elementary particles, but when these combined to form physical elements, stabilising the previous interactions, new interactions emerged between the elements themselves. And so on up through the realms. Biological interactions displaced those from which life emerged. Mental interactions displace those of a purely biological nature. And virtual world interactions supersede those in the real social and physical realm.
As increasing information interactions take place between linked computers dispersed around the globe, there emerges a new structure that becomes the agency for further interaction. The phrase ‘the network is the computer’ takes on full meaning. While current agency resides in individual humans and their individual computer agents, the interconnected and distributed nature of information interaction today, and even more so tomorrow, will lead to collaborative agencies that provide added value through their greater reach. There is but a small step from there to an autonomous agency at a global level.
An instructive metaphor is the human brain, which has no little homunculus within it to direct operations, but which rather collectively brings to bear information from its various areas to create an agency that we then name our self and that guides our thoughts and actions. The same applies to our living body, all of its parts working together in harmony, in interaction with each other and without a master agency directing them individually.
That a global information processor based on the multitude of interconnected computers will emerge as an information agency is not surprising. And it need not cause concern if it is not confused with the kind of big bad controlling brain that is depicted in science fiction to wreck havoc and to be defeated by heroic humans.
Such an emergent structure, built from the myriad interconnected local information processors, will simply guide further knowledge creation through ever more sophisticated information interactions. Information autonomy then becomes relocalised in this aggregate super-structure, beyond all its individual contributing elements, where it currently resides.
4.3 Unbounded autonomy.
The ultimate question of course is whether information interactions can take place without the support structures provided by the previous realms. Can they become fully disembodied?
Philosophically, that would mean becoming pure process. It would mean becoming pure transition between structures, a form of energy that is for a time unbound, without a body that captures and holds it for later deployment. A tall order, conceptually!
And yet, our discussion earlier of information as a signal has that needed intangibility to grasp this possibility. In the immediate, as it is occurring, a signal is disembodied. It may be carried on a medium (the sound of thunder being a wave within the air, for instance), but yet its significance as signal is well beyond that. Perceiving and interpreting the signal has little to do with the physical medium.
Until information gets stored (in a brain, a book, a computer...), it is in processing, in a disembodied state whose unfolding has little to do with physics. Rather, it has all to do with logic, the logos that underlies the coherence of this world. And if we follow Leibniz in his belief in a logical calculus that, given enough information, can map out all consequences, then, the information spiral in the making will inexorably lead to a full representation of the world. This, of course, is highly speculative, but remains plausible.
Such a representation would constitute an abstract structure that maps out the world and processes it through information interactions. Just as the laws of nature are not captured anywhere except in nature itself (in its structure and operation - its unfolding), so too information interactions will be embodied in the world as a whole rather than being coded in some medium within the world. Information will become the world, just as physics became the world soon after the initial big bang.
Computing, as information processing, is a manipulation of abstractions that used to be the sole privilege of the brain. Today, it is also performed by artefacts external to the brain, namely computers, which engage in effect in artificial cognition. While all realms remain grounded physically, we see a breaking of the prior-realm grounding principle in the development of artificial cognition. Computers are not grounded biologically. They are cognitive beings (defined on that plane of operation) which are constrained physically, but not biologically.
The question that naturally surfaces at this juncture is whether we might witness yet a further instance of the principle of prior-realm breaking, with computing distancing itself this time altogether from the physical realm. In what cognition would then be instantiated (disembodied, but nevertheless operative) remains an open question.
We have reached a point here that attains the borders of our current capacity of comprehension. We might see the road that is followed, without being able to follow it ourselves. But just knowing of that road will inform our current view of our own cognition and of the business of cognitive ergonomics.
Dennett, D., 1996. Kinds of Minds: Towards an Understanding of Consciousness. New York: Basic Books.
D’Inverno, M. & Luck, M., 2001. Understanding Agent Systems. Berlin: Springer Verlag.
Donald, M., 1991. Origins of the Modern Mind. Cambridge, MA: Harvard University Press.
Duchastel, P., 2001. Towards an Information Realm. Digit-HUM, 3, [Online] http://turquia.uoc.es/humfil/articles/eng/duchastel/duchastel.html
Duchastel, P., 2002. Beyond HCI - towards Information Interaction. Submitted for publication.
Levy, P., 1998. Becoming Virtual: Reality in the Digital Age. New York: Plenum.
Mainzer, K., 1997. Thinking in Complexity. Berlin: Springer.
van Dijk, J., 1999. The Network Society. London: Sage.
Woolridge, M., 2002. Introduction to MultiAgent Systems. New