Brian R. Gaines and Mildred L. G. Shaw
Knowledge
Science Institute
University of Calgary
Alberta, Canada T2N 1N4
System-theoretic accounts of the epistemological processes underlying knowledge acquisition have been shown to apply to both individual human behavior and social development processes, and to enable algorithms to be developed for computer-based systems modeling. Such accounts are applicable to the upper levels of the hierarchy of autonomous systems to provide models of socio-economic behavior. In this paper they are applied to the development of information technology, and used to account for past events and predict future trends in relevant industries such as computing and genetic engineering. Underlying all developments in information technology is a tiered succession of learning curves which make up the infrastructure of the relevant industries. The paper provides a framework for the industries based on this logical progression of developments. It links this empirically to key events in the development of computing and genetic engineering. It links it theoretically to a model of economic, social, scientific and individual development as related learning processes with a simple phenomenological model. It uses this model to account for past developments in information technology and extrapolates it to predict future trends.
Advances in computing science and molecular biology are two of the most important phenomena of our age. They are each representative of the development of information science as a new foundation for human knowledge. They are both predictive sciences that can model natural processes: the reasoning and evolutionary processes of living systems, respectively. They are also both expressible operationally as normative sciences that can generate new technologies that modify and go beyond natural processes: electronic computing to extend reasoning and genetic engineering to extend evolution, respectively. At a fabrication level the two technologies have much in common: microelectronic device manufacture is the control of information structuring semiconductor material at a molecular level to create non-living knowledge processors; genetic engineering is the control of information structuring organic molecules at a molecular level to create living systems. Together the two technologies illustrate the massive extent of information science, encompassing the furthest reaches of human rationality and the ultimate foundations of life, and bringing both into the technological domain, subject to human control and design.
Many writers have speculated upon the impact of computing and genetic engineering on our society, culture and value systems (Wiener 1950, Bell 1973, Mowshowitz 1976, Weizenbaum 1976, Carney 1980, Cherfas 1982). The impact is now being felt and many commentators are attempting to analyze its nature and project its trends (Toffler 1980, Dizard 1982, Krimsky 1982, Brod 1984, Turkle 1984, Glover 1984, Nossal 1985). However, the majority of such speculation and analysis offers one-sided perspectives relative to two key distinctions. The first perspective assumes that the major impact is of information technology on society rather than society on information technology. This takes for granted Ellul's (1964) view of technology as autonomous, and fails to question why the technology itself has arisen out of the processes of society. The second perspective assumes that the future may be projected as an extrapolation of processes begun in the past. This takes for granted the causal models fundamental to physical science and fails to recognize the higher level goals that lie behind information technology developments seen as part of the teleology of society.
These one-sided perspectives are valid and lead to meaningful conclusions relative to their presuppositions. They are appropriate to the analysis of technological dependencies for purposes of industrial and government planning, for example, that one must have adequate capabilities for semiconductor device manufacture or DNA cloning before one can develop computer or genetic engineering industries. However, they are not appropriate perspectives from which to evaluate the future development of information technology which involves relations between information processing, society, values, culture and creativity. These last four are phenomena of the life world (Schutz & Luckman 1973) and embedded in its processes. To understand their relations to information technology we must view it also as a phenomenon of the life world embedded in its processes, both generated by them and generating them (Blum & McHugh 1984). If we treat technology as autonomous we forget its roots:
"Technology is the human's achievement, not his failing--even though the use he chooses to make of it may be fallen indeed. If the products of human techne become philosophically and experientially problematic, it is, I would submit, because we come to think of them as autonomous of the purpose which led to their production and gives them meaning. We become, in effect, victims of self-forgetting, losing sight of the moral sense which is the justification of technology. Quite concretely, the purpose of electric light is to help humans to see. When it comes to blind them to the world around them it becomes counterproductive. The task thus is not to abolish technology but to see through it to the human meaning which justifies it and directs its use." (Kohak 1984)
Information technology is directed by the processes of society at least as much as it impacts them. Up to its present stage its development has been governed by existing social processes and the dominant value system. It is only just beginning to become an agent for social change. The perspective that views only the impact of computing on society misses the major dynamics of information technology during the past forty-five years. In particular, technological forecasting is unlikely to be a basis for any except very short-term projections of the future of computing and genetic engineering. The social objectives that determine the economic pressures on these industries are satisfied by selecting from the pool of available technology of the most effective technologies to satisfy market requirements. The inertia of social development is very much greater than that of technological change and long-term forecasts are most accurately based on social objectives rather technological extrapolation.
Viewed as a causal chain the development of information technology is highly indeterminate. Technological forecasts for computing are notoriously difficult (Gaines 1984a), and the younger industry of genetic engineering has developed at a startling pace (Cherfas 1982). Projections of the technologies over more than a few years are meaningless. Each month the technical press contains a number of major surprises. The causal model is not appropriate to a phenomenon governed by social needs and hence directed by the goals of society. The past is not an adequate determiner of the future in these circumstances. A teleological model that views society as moving towards some future situation is more appropriate to the processes of the life world. The cascade of sub-goals generated in order to create that future situation accounts for the activities of the past, generates those of the present and determines those of the future. The determinacy, however, is that of a goal-seeking, adaptive system. It predicts the necessary achievements but not the means by which they will be achieved.
"The world is overwhelmingly complex for every kind of real system... Its possibilities exceed those to which the system has the capacity to respond. A system locates itself in a selectively constituted `environment' and will disintegrate in the case of disjunction between environment and `world'. Human beings, however, and they alone, are conscious of the world's complexity and therefore of the possibility of selecting their environment--something which poses fundamental questions of self-preservation. Man has the capacity to comprehend the world, can see alternatives, possibilities, can realize his own ignorance, and can perceive himself as one who must make decisions."
This model seems to underlie De Bono's (1979) views of the role of computers:
"By great good fortune, and just in time, we have to hand a device that can rescue us from the mass of complexity. That device is the computer. The computer will be to the organization revolution what steam power was to the industrial revolution. The computer can extend our organizing power in the same way as steam extended muscle power... Of course we have to ensure that the result is more human rather than less human. Similarly we have to use the computer to reduce complexity rather than to increase complexity, by making it possible to cope with increased complexity."
Wojciechowski (1983) sees the coupled dynamics of complexity processes and information technology as the most recent manifestation of the growth of the knowledge construct in human society, the infrastructure of information acquisition, storage, processing and dissemination that is part of our culture and supports individual and social knowledge processes. He has developed an ecology of knowledge and proposed a set of laws underlying the dynamics of knowledge processes and their role in society (Wojciechowski 1986). He notes that, while attempts at complexity-reduction are a major social dynamic, the overall complexity of the life-world is increasing and that most human problems are now humanly generated:
"Contrary to past epochs, from now on the future of humanity, and ideed the very survival of the human race, is in its own hands. Man will have a future if he is capable of envisaging a positive course for himself and planning for it. The capacity of dealing with the future entails the ability to cope with the present and future problems." (Wojciechowski 1986)
The conceptual framework established by Luhmann and Wojciechowski provides foundations for a model of information technology and its role in society. This paper gives substance to these models of society and information technology by extending the notion of a personal scientist modeling the world (Shaw 1980) to that of a society learning a technology, a communal scientist (Shaw & Gaines 1986). This enables Marchetti's (1981) concept of society as a learning process to be used in order to model the development of information technology:
"The concept of a learning society, with its implications on the ecological Volterra equations, represents a very powerful tool in organizing social behavior and hints to the possibility of a unified theory for genetic evolution, ecology, sociology and economics."
In analyzing the architecture of modeling systems, Klir (1976, 1985) proposed an epistemological hierarchy accounting for the main components of any modeling systems and their inter-relations. Gaines (1977) gave a mathematical formulation of the general problem of modeling as a relation between order relations at different levels of the hierarchy. Gaines & Shaw (1981) gave a general interpretation of this hierarchy as generated by levels of distinction and showed how it was instantiated in psychological terms. The hierarchy has proved a valuable conceptual tool in analyzing a wide variety of modeling systems both in terms of their ontological presuppositions and their epistemological processes.
The notion of a distinction itself is a primitive concept underlying the representation of knowledge (Gaines & Shaw 1984b, 1985). It is a sufficient primitive to give foundations for systems theory including the notion of a system itself (Gaines 1980). In its psychological form, as a personal construct (Kelly 1955, Shaw 1980), the notion has been used to derive very effective techniques for knowledge transfer from experts to expert systems (Shaw & Gaines 1983a, Boose 1985). In the context of the development of information technology, the fundamental role of distinctions as determining the nature of knowledge is best put by Brown (1969):
"The theme of this book is that a universe comes into being when a space is severed or taken apart...By tracing the way we represent such a severance, we can begin to reconstruct, with an accuracy and coverage that appear almost uncanny, the basic forms underlying linguistic, mathematical, physical and biological science, and can begin to see how the familiar laws of our own experience follow inexorably from the original act of severance."
Their foundational role in knowledge acquisition is evident in the hierarchical representation of distinctions in a modeling system shown in Figure 1. It is important to note that this hierarchy does not introduce any additional primitives beyond that of making a distinction. The levels of the hierarchy are the results of distinctions that we make. Thus, in Klir's (1976) terminology:
Note that the upper levels of modeling are totally dependent on the system of distinctions used to express experience through the source system.
Figure 1 Epistemological hierarchy of a system modeling a world
Distinctions are not just static partitions of experience. They may be operations: actions in psychological terms; processes in computational terms. Whether a system finds a distinction in the world, imposes it passively as a view of the world, or imposes it actively as a change in the world, is irrelevant to the basic modeling theory. It makes no difference to the theory whether distinctions are instantiated through sensation or action. We can make a distinction passively or actively. We can sense some condition already in the world or we can act to establish some condition in the world. In system-theoretic terms there is no intrinsic distinction between prediction and control. In scientific terms the predictive goals of scientific investigation and the normative goals of technological change are basically indistinguishable. In biological terms, a living system may discover a comfortable niche, create one, or combine the two processes.
The hierarchy of Figure 1 accounts for learning processes as the modeling of events enabling adequate prediction and action. A modeling schema results from distinctions about distinctions at each level in the hierarchy. In prediction the key distinction is to what degree a level accounts for the information flowing through it and hence this distinction may be termed one of surprise (Gaines 1977), in the sense used by the economist Shackle (1955). Surprise goes in opposition to the degree of membership (Zadeh 1965, Gaines 1983) of a predicted event to an actual event and the expected surprise is a form of entropy. Surprise at the lowest level of the hierarchy corresponds to distinctions being inadequate to capture events; surprise at the next level to inadequate variety to experience events; at the next level to inadequate approximation to predict events; at the next level to inadequate simplicity to explain events; at the next level to inadequate comprehensiveness to account for events.
The formal theory of modeling is one in which models are selected at each level down the hierarchy to minimize the rate at which surprise is passing up the hierarchy. The criteria for model selection independent of the data are generally thought of as being ones of simplicity/complexity: of two models which fit the data equally well choose the simplest. However, notions of simplicity/complexity are not well-defined nor intrinsic to the class of models. The simplicity/complexity ordering is arbitrary and in its most general form is just one of preference. Hence the general modeling schema is one in which surprise flows up the hierarchy and preference flows down. In situations that are mathematically well-defined, such as determining the structure of a stochastic automaton from its behavior, such a model schema gives the correct results (Gaines 1977). Conversely, the success of the schema in stabilizing with regard to a given world defines the characteristics of that world.
Thus the basic modeling schema for learning from experience is one in which surprise flows up the hierarchy and preferences flow down. In primitive organisms only the lower levels of the hierarchy are developed, surprise is generated from experience and preference is genetically encoded. In higher organisms the modeling process generalizes both surprise and preference to cope with novel environments. Human life has developed the upper levels of the hierarchy and detached surprise from experience and preference from its genetic roots. Surprise can flow up from a level without flowing into it from below because the processes at that level have generated novelty. Preference can be generated at a high level detached from both experience and genetic origins and flow down to affect the relations of the organism to the world.
The loop in Figure 1 from events through distinctions up through the modeling hierarchy and then down again to predictions and actions characterizes what Shaw (1980) has termed the personal scientist. This derives from the epistemological model of man as an anticipatory system developed by Kelly (1955) as a psychology in which the individual modeling the world is seen as man the scientist. Kelly's personal construct psychology (PCP) provides an extensive theory of both normal and abnormal human functioning, which has strong systemic foundations (Gaines & Shaw 1981) and has been operationalized through computer programs (Shaw 1980). PCP models a person as making distinctions about experience termed personal constructs and described as:
"transparent templets which he creates and then attempts to fit over the realities of which the world is composed" (Kelly 1955).
A person's construction process is a basis for anticipatory and fallible knowledge acquisition:
"Constructs are used for predictions of things to come, and the world keeps rolling on and revealing these predictions to be either correct or misleading. This fact provides a basis for the revision of constructs and, eventually, of whole construct systems." (Kelly 1955)
The systemic hierarchy of Figure 1 has an analog in psychological terminology as shown in Figure 2. The source level is one of constructs, distinctions made in interacting with the world. The data level is one of experiences, events which happen to us, and we make happen, in terms of the distinctions already made. The generative level is one of hypotheses which are rationalizations of experience. The structure level is one of analogies which are correspondences between these rationalizations. The meta level is one of abstractions which are foundations of analogies. The meta-meta level is one of transcendencies which are accounts of abstractions. Interaction with the world is, therefore, mediated through the construct system to produce experience which is modeled through the higher levels and leads to predictions, decisions and actions again mediated through the construct system.
Figure 2 Construction hierarchy of a person modeling a world
Kelly (1955) places the major emphasis of his work on the notion of constructive alternativism, that we have a choice in our construct systems at every level in the hierarchy and that real-world problems may often be solved by exercising this choice. Note that this should not be interpreted as an idealist position that ascribes all phenomena to our interpretation of them. Since the construct hierarchy also leads to decision and action, changes in it may equally affect the real world. Kelly and Brown are both neutral to any philosophical stance such as idealism versus realism; it is the distinctions which a philosopher makes that determines his stance and these can be analyzed in terms of the modeling hierarchy. Kelly saw his theory as reflexive and the only fundamental principle, apart from that of a construct itself, being that of constructive alternativism.
The anticipatory processes of the modeling hierarchy may be extended to the operation of society by viewing groups of people as larger cross-sections comprising multiple individuals (Shaw & Gaines 1981). This concept may be given deeper significance by considering the inductive inference process underlying knowledge acquisition and modeled in the hierarchy. Whereas the deductive logical inference that underlies the operation of conventional computers is well-understood and well-founded, the inductive inference that underlies human learning is not. Deduction guarantees to take us from valid data to valid inferences, but the inferences are thereby part of the data--no new knowledge is generated. Induction takes us from valid data to models of that data that go beyond it--by predicting data we have not yet observed, and by giving explanations of the data in terms of concepts that are unobservable. Induction generates new knowledge but, as Hume (1739) pointed out over 200 years ago, the process is not deductively valid and it is a circular argument to claim that it is inductively valid.
Philosophers have continued to debate Hume's arguments and search for justification of the inductive process. Goodman (1973) proposed that we accept the circularity but note that it involves a dynamic equilibrium between data and inference rules as shown in Figure 3: "A rule is amended if it yields an inference we are unwilling to accept; an inference is rejected if it violates a rule we are unwilling to amend." Rawls (1971) in his theory of justice terms this a reflective equilibrium. Recently Stich and Nisbett (1984) noted flaws in Goodman's argument and repaired them by proposing that the equilibrium is social not individual: "a rule of inference is justified if it captures the reflective practice not of the person using it but of the appropriate experts in our society." This argument arose in the context of the explanation of the authority of experts in society, but it is also significant in suggesting that the basic system underlying knowledge acquisition has to be taken as a society rather than an individual.
Figure 3 Reflective equilibrium in inductive inference
The extension of the modeling hierarchy to social processes is straightforward since Figure 1 presents a general modeling schema and applies as much to groups of people, companies and societies as it does to the roles of a person. The epistemological hierarchy of a person is a cross-section of the epistemological hierarchy of the society generating their life-world. Pask's (1975) concept of P-Individuals as the basic units of psycho-socio-processes allows roles, people, groups, organizations and societies to be treated in a uniform framework (Shaw & Gaines 1981, 1986). An individual is defined in cognitive terms as a psychological process (Pask 1980) and more complex psychological and social structures may be defined similarly by taking into account the possibilities of timesharing, process switching and distributed processing with psychological processors. For example, one person may assume many psychological roles (process switching), whereas a group of people working together may act as a single goal-seeking entity and hence behave as one process (distributed processing).
In the analysis of technology the skills to achieve goals in the world are the crucial capabilities of the modeling system. Figure 4 shows the basis for action at different levels in the modeling hierarchy.
Figure 4 Action hierarchy of a system modeling a world
It is an interesting comment on the state of the art in computer science that it has proceeds "middle-outward" in its representation of the knowledge involved in skills at different levels of the hierarchy. Information technology has been primarily concerned with level three activities, and is only now beginning through expert system developments to emulate level two activities. The primitive sensation-action modes of learning at level one require the large-scale parallel processing essential to the emulation of human vision and locomotion, and will be developed as part of the next generation of robotic systems. The higher level functions of levels four and five are being studied in artificial intelligence but require developments in mathematics and system theory for their full realization.
The creation of new knowledge takes place through the surprise/preference flows within the hierarchy and it is these processes that determine the rate of technological invention and product innovation. The human capability for an entire society to act as a distributed knowledge acquisition system is dependent on the role of communication processes in coordinating activity at a given level of the hierarchy across different people. This communication process whereby each person does not have to undertake all aspects of the inductive process but can share the results of such processing by others supports what is generally termed the culture of a society. People use language for much of this communication but they also have in common with other animals the capability to communicate cultural information without the use of language. Mimicry is an important mechanism for knowledge acquisition as is reinforcement through reward and punishment.
The human development of language enables coordination to take place in a rich and subtle fashion that greatly enhances, but does not replace, the more basic mechanisms in common with other species. It is particularly important at the upper levels of the hierarchy where direct expression is difficult. From an individual point of view, language is a way of by-passing the normal modeling procedures and interacting directly with the system at any level. In particular it can directly affect the preference system. Even when language cannot mediate the direct transmission of knowledge it may be used to achieve the same effect by the indirect support of other mechanisms, for example, one can describe a good learning environment, or a behavior in sufficient detail for mimicry. Language is essential to much of human learning, and our interaction with the knowledge construct (Wojciechowski 1983, Gaines & Shaw 1983b) is just as important as our interaction with the world (Shaw & Gaines 1983b, 1984, Gaines & Shaw 1984a). The evolutionary pressures would be very strong in selecting genes giving the capability for a species to act as a single distributed individual, combining autonomy and cohesion through enhanced linguistic communication. Linguistic transfer of knowledge is the most important process for the dissemination of information, for example in technology transfer.
Figure 5 shows the cultural support for knowledge acquisition at different levels in the modeling hierarchy.
Figure 5 Cultural transmission hierarchy of people modeling a world