by Hugh M. Lewis

Alternative systems are in a sense a logical outcome of the development of human cultural systems--they are an extension of the basic constructive capacity of humankind. Humans have, by virtue of the application of their natural symbolic intelligence, become capable of creating entirely new systems that were previously unprecedented in the natural scheme of things. Generally we refer to these as cultural artifacts, and we have many examples from the earliest periods of Hominid evolution--stone tools, hearths, clothing. The rise of technological and industrial civilization has born witness to the proliferation of artificial systems that exist in reality because they have been made by people, and follow logical and scientific principles, but which have no known natural antecedents or correlates. The rise of alternative, humanly constructed systems has achieved a basic augmentation of reality, and of our knowledge systems that are conjoined to reality, in many different ways. The hydrogen or atomic bombs were but science fantasies until their first demonstrations on nuclear testing grounds. Similarly, human flight was believed impossible until the advent of the Wright Brother's new engine-powered plane. As a result of these basic advances, our world has been transformed rapidly and irretrievably in many ways.

There is the sense emergent from the history of the development of alternative systems that such development eventually follows a certain course of inevitability--human intelligence is curious, and it is questioning and exploring in basic ways. Eventually, humankind would hit upon basic solutions to basic scientific problems, and these solutions would in time facilitate the integration achievable by such systems and lead in time to the acceleration of the pace of such technological development and integration as we have witness in the world. Such development would proceed much more rapidly if it were not for some sense of historical inertia from cultural and structural frameworks of human society that serve to interfere with and impede such development. There is a sense, in other words, that "if something is logically possible, then it is eventually inevitable."

Humans were meant to fly, not because they were graced with wings, but because their brains allowed them the imagination to conceive of such flight, and the determination to try to fly in whichever manner seemed possible. The possibility of alternative systems development therefore lies latent in the ground of natural systems patterning, to be discovered like a gem in the rough, by some errant explorer.

There should be therefore several expectable trendlines forthcoming in the progressive development of alternative systems:

1. As sophisticated solutions to complex problem sets, these alternative systems should become more streamlined such that they represent the best possible fit for the problem set they are designed for.

2. As mechanical systems, these alternative systems should be based upon the ability to create and manipulate increasing levels of power and energy, defined in the classical sense as the ability to do work.

3. Also as mechanical systems, these alternative systems should develop increasing levels of efficiency in terms of the ratio of work to total energy involved in the process. Not only will such systems be capable of working harder to do more work, but the work that such systems do will be both more efficient and more suitable as a solution to the problem posed for the work in the first place.

4. As systems involving energy dynamics and exchange, alternative systems are also information-based systems. In other words they rely upon the organization of information within and by such systems to achieve a degree of functional order. From a standpoint of general problem solving, such systems can be said to be "intelligent" in that they are capable of autonomously tackling problems of some level of sophistication.

5. As systems evolve and develop in relation to different problem sets at different levels, alternative systems should demonstrate two interrelated trendlines for development. First, they should become increasingly integrated between different levels and different areas. Secondly, they should become at the same time increasingly generalized and/or specialized in their relation to particular or generalized problem sets. In other words, in relation to this final point, it can be said that systems do not evolve in a theoretical or practical vacuum of application, but in the context of the development of many other alternative systems at the same time. Thus, the term "alternative systems" has implicit to its name the general context in which such systems arise and co-develop in the first place. It is evident that as alternative systems develop, previous systems will give rise to newer and newer systems, and this process should in time increase in its rate and amount of production of new alternative systems.

The increasing degree of integration of such systems determines in part that, as they develop, it will become increasingly difficult to differentiate and clearly separate where one kind of system leaves off and another kind takes over. In general, it can be expected that alternative systems should grow in complexity, sophistication and power to greater and greater levels that are integrated. It is expected therefore that such systems should exhibit a form of stratification of function at different levels and in different ways.

One constraint that appears to be operative of all such alternative systems is the notion that such systems are ultimately designed for and based upon human systems, and that they serve in one way or another the basic problems of adaptation and successful integration of the latter systems. The notion that alternative systems could arise that are essentially self-serving in function and independent of the human designers or designs, is one that is not uncommon in science fiction. Alternative systems, as integrated supersystems, have not yet reached such a stage of independent functioning in the world. Furthermore, it is a clear case of interdependency between human and alternative systems, such that alternative systems depend upon the human designers and operators for their functioning as much as humans have come to depend upon such systems for their functioning and operation in the world. I do believe that in time, alternative systems will become increasingly autonomous in function from the design influence or control of human manipulators of such systems. Thus I would append a sixth point to the five above, and claim the following kind of trend:

6. Alternative systems should develop in a manner that is increasingly but relatively autonomous at more and more points of human control functions, such that human design and control in such systems will play a diminishing part in the overall functioning of such systems, restricted perhaps to their initiation, original construction, monitoring and possible repair. Even these functions may eventually be taken over increasingly by the design of alternative systems.

Ultimately, at some point, one would expect that such systems would incorporate a degree of self-design and self-determination of patterning that is more or less completely separate from the role of the human operators or designers in such systems. It is doubtful that such systems would achieve a degree of automation of pattern that is completely independent of human design or control, but it can be imagined that such human influence will diminish as systems increase in intelligence and sophistication. It is doubtful furthermore that humans want or require such systems to be completely autonomous, as their essential purpose is to extend and elaborate the function of human systems in the first place.

Furthermore, all alternative systems are by definition human designed systems, and thus they are systems that have inherent to their existence the functioning of human systems and human designs. Such systems will achieve relative autonomy only to the extent that humans are capable of designing a sense of autonomy into such systems in the first place. All such designs are by definition explicit systems defined by relational rules and object-values--in other words, it would be difficult to confer upon any alternative system qualities or properties of behavior that are difficult in themselves to define or describe in any systematic manner. By definition, with the state of hardwired intelligence as exists today, all such relationships would be, furthermore, strictly logical in pattern, though it is clear that natural intelligence does not always function in a purely or strictly logical manner. A consequence of this is the constraint of the anthropological relativity of such systems. Such systems would have no intrinsic duality of patterning in its signification structures--in other words they lack any implicit level of intuitive meaning that comes attached to the signal. It is not to say, like the classic ELIZA example, that humans cannot be tricked into believing that computers have such an implicit level of meaning, and this kind of parlour trick constitutes in essence the basis for hard A.I. criteria for intelligence in the first place. It must be reiterated, that such implicit, intuitive meaning is totally lacking and impossible in artificial intelligence, and it represents an insuperable constraint of such systems to function as if they were natural.

*****

Part of the problem of the definition of alternative systems is defining the term intelligence in an operational sense that gives it significance to describing and explaining the role and function of information processing that sophisticated and semi-autonomous machines in the future will perform. If we define as "intelligent" the ability to solve complex problems, then the issue is more straightforward than if we attribute to the notion of intelligence some sense of autonomous sentience and self-awareness as a being in the world. Surely all forms of natural intelligence can be attributed such a relative kind of autonomous self-sentience that is unique to such systems. This is a principle attribute that can be used to separate natural from artificial forms of intelligence. However sophisticated seeming an artificial intelligence system may be, we do not realistically attribute true or real sense of autonomy or self-functioning sentience to such systems. They are nothing more than sophisticated appliances that function because they have been plugged in and turned on. Their intelligence is inherent to their design and functioning, and this has been purposively created by its human engineers.

This question brings to bear what criteria we adopt for measuring or determining what constitutes intelligent systems--in general we adopt a "hard AI" Von Neuman criteria of intelligence as resembling in everyway possible that functioning and response patterning of a human being may be seen as unrealistic in consideration of more limited models or criteria of artificial intelligence that seeks to solve specific kinds or sets of complex problems efficiently and reliably. In other words, a "soft" or functional AI approach is not only more sufficient, but therefore preferable to the challenge of development of forms of alternative intelligence in systems that can function autonomously or in an integrative manner.

The adoption of criteria of specific or general problem solutions, by design and control functioning, of complex problem sets or classes of problems, provides us with a more realistic approach to the challenge of the problem of artificial intelligence than we have with a "Chinese room" set of criteria. Alternative intelligence does not and cannot resemble forms of natural intelligence except in a superficial analogical manner--it lacks the direct organic sense of its history and evolutionary contexts of development. The manner that it achieves solution to problems, however defined, is fundamentally different from the way that natural brains solve problems in the real world.

Definitions of artificial intelligence that are part of the design of alternative systems are bound centrally by the problem of the anthropological relativity of all such designs.We cannot conceive of a form of intelligence which, by de facto conception, becomes bound by own own sense of human intelligence. The only time or possible scenario in which this will no longer hold true will be if and when we encounter forms of alien intelligence whose intellectual capabilities evolved differently and independently from our own. Alien intelligence would give rise to alternative systems that may in many ways resemble or be convergent with our own in the solution of common kinds of mechanical problems. At the same time, it is likely that such alternative systems may be completely different from our own in some essential ways, hence almost undecipherable as such in terms of our own design configurations and understanding.

The problem of anthropological relativity in relation to our conceptioning of systems of alternative intelligence is best exemplified, I believe, in our tendency to stereotype artificial intelligence systems with anthropomorphic forms and functions. Even hard AI criteria is based upon the principle of such an anthropomorphic stereotype, even if the best that can be achieved is mere an illusion, a parlor trick, a mimicry of real human intelligence. The greatest danger in the anthropomorphization of the function and role of artificial intelligence is, I believe, in unduly constraining our solutions to A. I. type problems, and even in our definition of such problems in the first place, in ways restricted to our preconceived notions about what human intelligence is supposed to be like. If Alternative intelligence in its design can be loosened from the anthropomorphization of such systems, then it becomes apparent that such systems can be applied in a very wide variety of ways in a variety of settings that do not require a human-type answer. We do not have to conceive of mobility of robotic systems as bipedal or even as with legs, whereas some kind of tracking system might represent a much more efficient and manageable solution to such problems of independent mobility.

*****

The conception of alternative systems I believe comes to focus upon the design of hybrid analog-digital types of computer systems that are capable of serving a wide-range of functions. Such systems should be distributed in function, interlinked in a network of multiple processing systems. We can see such systems also being distributed in terms of the numbers and kinds of inputs available to such a system. Such systems need to be capable of reading directly from fluctuations of natural signals in the environment and monitoring such systems without human mediation involved in the monitoring.

At the same time, such systems can also be distributed at the output level, and this output can consist of a variety of functions that articulate and manipulate environmental elements in basic ways.

We can also refer to distributed control systems that allow feedback to occur between different functional levels or components of different levels, as well as distributed input-output interfaces that control and manage human signal inputs and outputs in a manner that allows human manipulation and redesign of the system on one hand, and that can generate meaningful results that are digestible to humans, on the other. The internet offers the possibility of the construction of such a super-system, and such systems have already been engineered, albeit in yet rudimentary form. In the construction of such a system, input and output components can be proximate or remote from one another. The entire system could potentially encompass the globe, or even be extended beyond the boundaries of the earth to compose a solar satellite system.

It is evident that we can refer to a totally distributed system in which all these functions are spread out over a series of interconnected devices each capable of working both independently and in a manner that is coordinated with all the other systems. In such a system, there would be no central hub or main control component upon which all the other systems would depend. New systems can be added or modified at any point within the overall system. Furthermore, each unit or component of such a system could itself be functionally integrated to perform simultaneously any number of different functions in the overall system at the same time. This design of alternative intelligence systems must occur independently of the component design of the system, or of the nature of the design of the individual component. Hybrid systems span more than merely the digital and simple analog wave functions that are built in conventional types of systems, and can embrace a number of more exotic architectures as with quantum type computing or light computing.

Totally interconnected and distributed systems puts a premium upon the communicative function between components, as well as upon the distributed control systems that accompany each of the components and serve to integrate this component to the larger overall system. Such systems must be capable of finding the other components of the network, talking with these components in a meaningful manner, and then determining between themselves the kinds of decisions that have to be made that affects the outcomes and state-pattern of the system as a whole. It seems that to a great extent the invention of the internet has obviated this communication and control aspect of such distributed systems, though the state of development of the internet may just be reaching a level of sophistication to allow these kinds of advanced distributed systems to become a real possibility.

Such totally distributed systems can also be stratified upon different levels of function, and this stratification can be made into a form of self-organizing behavior of the system that is based upon its modular partitioning of functions to solve problem sets at different levels. Another aspect of this kind of system is provisioning the entire system with built-in memory functions that allow it to store different forms and levels of information and to be able to utilize this information in the solution of future problems. Memory storage would have to be itself stratified and organized functionally, as human memory is, and it would have to be construed as part of an active control system that allowed its continuous updating and re-envisioning.

Such systems would have to be working systems if they are to achieve their place in the world as alternative systems. They must perform useful functions, and in this the actual production of work is no less a part of such distributed systems than is the continuous monitoring of the environment. It is not difficult to imagine a number of different functions for such systems, in terms of education, entertainment, production work, environmental regulation, energy production and distribution, communication and even transportation systems.

Another aspect of such systems, I believe is that they achieve a degree of symbolic realism in their patterning of behavior that is not unlike that of human symbolic cognition. This is not to re-impose an anthropomorphic model upon such systems, or to adopt a version of the hard AI criteria. Rather I believe that especially in the interface requirements in the articulation of alternative systems with their human agencies, such symbolic patterning could prove to be quite useful and even powerful for augmenting the intelligence of the distributed system in ways involving pattern recognition and identification in symbolic terms. Symbolic criteria would allow us, at least in principle, to overcome partially the inherent constraint of artificial systems in lacking duality of meaning and an implicit level of signification. One consequence of this is in terms of genuine pattern-response recognition of natural forms in the environment, without this recognition being moderated or influenced by human mediators.

*****

The consideration of forms of alternative systems in relation to artificial intelligence leads to the speculation about some first principle of design that serves as an anchor point or key for the articulation of such systems in an effective manner. It may be that there are several interacting anchor points that are necessary for such a sophisticated system, though it is as yet not exactly obvious what such anchor points for such systems would be.

It is evident that such systems can be abstractly and mathematically defined with a great deal of precision and logical coherence attached to them. At the same time, they can be constructed in a manner as to yield fairly reliable and consistent results within limited empirical contexts of their articulation and interaction with the environment. One characteristic of such programming is its essentially linear character. Such programs function by processing single or multiple streams of information very rapidly, setting off in the process numerous secondary reactions and responses that may trigger even further patterns of processing. They apparently do not solve problems in the manner that human beings normally do, often basing final decisions on blind faith or at least strong intuition rather than on any strictly logical calculation. It is not clear either that multi-stream, parallel processing structures overcome this basic linearity of such systems.

*****

For the most part, the rise of alternative systems has been achieved in a piece-meal and uncoordinated, almost serendipitous manner. With the rise of the sciences, especially within the last two centuries, this process has become increasingly organized and deliberative, to the extent now that there are entire research organizations dedicated upon one level or another it the articulation and elaboration of alternative systems in reality.

Alternative systems offer to humankind one set of unique possibilities that are critical to the future of such systems and to human and biological systems on earth--alternative systems offer humankind the possibility of being able to transcend the natural constraints imposed by basic bio-geophysical systems, and to escape from the rule of the biological imperative that imposes the ultimately outcomes for all life on earth. Even now as we are learning through our alternative systems applications to manipulate the genome and reengineer genetic structures to our own designs, we are gaining rapidly a degree of leverage and control over biological processes and outcomes that were until now unprecedented. And yet we are failing in basic ways to achieve the full measure of control over our own human systems that is a necessary prerequisite to such successful application of alternative systems. As a result, much of our applied systems ends up being, in the structure of the long run, destructive and entropic in a manner that puts us closer to, rather than further from, our own basic biological imperatives.

All alternative systems that exist are essentially the product of human invention and construction. As such, it can be said that they are a product of human intelligence and its application to the manipulation of the environment. For the most part the motivations to do so stem, however indirectly, from the adaptive advantages such manipulations confer upon human beings. Without our intelligence, we could not have achieved such systems. In a very real sense, therefore, alternative systems are fundamentally cybernetic systems, if we accept the conventional definition of cybernetics as being the modeling of informational systems based upon natural forms of human intelligence. Even rather crude types of systems, for example the machine screw and the mouse-trap, are basic demonstrations of certain principles and patterns of intelligence. This is not just in their design and manufacture, but in their functional application.

It follows from this observation that the trend-line of alternative systems follows a gradient of increasing informational integration and capacity, to the extent that such machines will become increasingly intelligent in a cybernetic sense. I will not claim that machine intelligence is an inevitable outcome of the progressive development of alternative systems, but it seems to me that we would eventually hit on the right combination of things that would confer some nominal sense of intelligent functioning to a machine. With the rise of the information revolution, we are arriving at an age in which machines are becoming not just demonstrations of intelligent design and function, but coming to mimic and incorporate the principles and functions of intelligence into their very design and function.

In this regard, I believe, we must be careful to say that machine intelligence, or what I will refer to as alternative intelligence, is not the exact equivalent of native human intelligence, nor in this sense do I think it can ever become what can be called genuine or true intelligence, which the implications of self-awareness and sentience that we attribute as qualities intrinsic to true intelligence. What we have are extremely sophisticated machines, but machines which are non-biological and which do not heed any form of biological imperative. They are things that we turn off or on at our own choosing, and use as tools to our own ends. In this regard, the standard for artificial intelligence, which is human-like intelligence, and which is the goal of hard AI research, seems to me to be unrealistic and unlikely. This is not to say that we will not eventually build machines that have a sense of independent sentience in some manner, but when such a machine emerges from the laboratory and factory, it will not in any form but the most analogous resemble natural human intelligence.

In this regard also I must separate the question of artificial intelligence, as this is a focus of much cognitive science research applied in various ways, from the notion of a more general form of alternative intelligence, in which human-like artificial intelligence is just one possible variety. When we can accept an expanded definition of alternative intelligence, as being somehow a logical consequence of the development of alternative systems, then we can begin entertaining broader notions of what such systems might be like and the functions they might serve.

In this, there are two interesting and indirectly related issues. First, there is no reason that alternative intelligent systems need to be anthropomorphized at all, or in the robotic manner that they tend to be stereotyped and which appears to influence much research and development work in artificial intelligence that sets humanlike intelligence as its standard and goal. We can selective anthropomorphize functional aspects of such systems--for example hands, or eyes, or speech, but that does not mean that we have to also give such machines human form or functional anatomy. There is no reason a robot needs to be bipedal to be intelligent, if it is perhaps preferable to build a machine that can walk like an insect, for example. This issue of the anthropomorphization of alternative intelligence has greater impact and influence on our models and agendas in this regard than we may realize or want to acknowledge, and I believe it tends to constrain our imagination as to what might in fact be possible with intelligent machining.

The other side of the coin of this issue is the observation that I've had since being involved in artificial intelligence research, and this is that the interdisciplinary aspects of cognitive science work is primarily psychological, computer and philosophical, and it tends to leave out entirely as an objective problem the issue of the social organization and articulation of knowledge, as a shared and intrinsically social phenomena. This derives in part I believe from a theory of language that is primarily psycho-linguistic and psycho-genic in orientation, to the exclusion of the communicative aspects of language as a social phenomena. If we refer to intelligent machining as a form of informational system, to which we can apply information theory, then we must also recognize that as informational systems, such forms of intelligence are also ultimately systems of communication to which we can apply communication theory and design principles as well. And this takes us back once again to fundamental arguments concerning natural human intelligence as well--as symbolic manipulating systems, the human brain is context bound to cultural frameworks of articulation within which it achieves integration.

We have become culturally dependent creatures in ways we scarcely realize and refuse to admit. Human intelligence was a functional product of the elaboration of human language as a linguistic system. Another way of construing this issue is to recognize that our intelligence is tied symbolically to larger knowledge systems, mostly that exist in the form of recordings as part of a common stock of knowledge. Knowledge is distributed and organized within a social landscape, and finds its articulation within this landscape. It follows therefore that intelligence, being based upon human knowledge systems, is fundamentally dependent upon these knowledge systems in a manner that is fundamentally external to the function of intelligence itself. Intelligence bereft of its knowledge context is like an empty machine with no signals to record. It remains merely a sophisticated device without significant content. It follows also that a great deal of knowledge and information processing that passes for intelligence comes prepackaged and pre-integrated as such. It comes as something preprocessed in an intelligent manner, compatible to its digestion by an intelligent system. These are inseparable parts of one and the same problem.

Another way of saying that is that, if we over-anthropomorphize our conception of artificial intelligence in terms of its possible designs and applications, then we also simultaneously fail to anthropologize more realistically or systematically our understanding of what both natural and artificial intelligence really represent, as an extension of our own knowledge and communication systems.

At the same time, we must also come to a realization that in nature there are other alternative forms of natural intelligence that deserve analysis and modeling in terms of intelligent systems, and that probably, in the grandest scheme of things, there are in the universe other forms of natural intelligence that are essentially non-human and that would constitute the basis for the development and introduction of entirely new forms of alternative systems than those that are humanly constructed and conceived. We might say that any true intelligence must, to so qualify, at least have the symbolic structure and function that we so readily recognize in human intelligence, but we cannot necessarily say what kind of symbolization this might entail. If we watch a dog jerking its legs in a puppy dream, we realize that dreaming is a fundamental aspect of the rise of a complex brain that can be said to have some level of sentient intelligence. Would we say, therefore, for instance, that some form of alien intelligence would necessarily have to dream in the same way that many mammals appear to do?

My central concern therefore in the elaboration of alternative systems theory is the elaboration of a more general and comprehensive model of what can be called alternative intelligence, and then of possible applied forms of this model of intelligence in ways that have some degree of efficacy as technological systems. Primarily I am concerned with hybrid forms of computer designs that combine digital with analog varieties of processing. I have been particularly interested in the use of light modulation and recording/recognition as the basis for the storage, manipulation and retrieval of patterned information. This is furthermore augmented by various network designs of distributed multi-processing systems and also of systems of intelligent environmental monitoring and articulatory response and manipulation. The model I have come up with resembles something more like an octopus than a human, though I do not thing that such alternative systems need to be constrained by an predetermined stereotypical form.

In this regard, I make a critical distinction in alternative intelligence between what I refer to as abstract intelligence, and what I would refer to as natural forms of intelligence. The latter represents those expressions of intelligent functioning that arise naturally as a result primarily of higher order brain functioning in biological organisms. A claim could be made that genetic systems constitute a form of intelligence, but I would claim it constitutes an intelligent system of informational patterning and organization, rather than a system of intelligence. This notion of natural intelligence obviously embraces the range of mental fluctuation of most primates, dolphins and many other large brained species. Complex problem solving has been demonstrated, for instance, in octopus, and even dogs show at times remarkable feats of memory, humanlike emotion and intuition, and even a rudimentary form of symbolic behavior and problem solving.

I deliberately contrast the notion of natural intelligence from the conception of abstract intelligence, per se, to highlight what I consider to be a critical difference and impasse between natural and artificial forms of intelligence as these latter have been invented by humankind. Natural intelligence arises biologically, based in the brain of the organism, and its rules and order of patterning remain mostly implicit to its functioning and behavioral consequences in terms of adaptation, communication and complex social relationships. Abstract intelligence arises primarily as a mathematical possibility, as a kind of formal-functional capacity of systems design that can be logically and parametrically ordered. All intelligent machines so far produced by human beings have exhibited mathematical abstract intelligence--not one of these has demonstrated the qualities of even the most primitive forms of natural intelligence. In general, rules governing the ordering and operation of abstract systems of intelligence are either formally or functionally defined in explicit ways. Its functioning does not arise as an emergent property of the integration of a complex organ, rather its sense of intelligence arises as the result of the top-down planning of complex systems based upon the reiteration or recursion of explicit and logically correct functions.

To push the distinction one step further, I would claim that the results and outcomes of a system of abstract intelligence are fundamentally different than those for a natural system of intelligence. Abstract intelligence achieves solutions to complex puzzle-problems that have a form of correct solution possible. Such problems may be astronomically complex, and there may be more than one solution pathway through the search solution space, but in general a solution to such a problem solves what is known as the Van Neuman bottleneck of an explosion of possible wrong solutions that develops when the problem remains unsolved. By contrast, I would claim that natural intelligence in general solves "problems" of an entirely different kind or quality. Generally, the problems solved by natural intelligence are those kinds off dilemmas to which there is no necessary single or even multiple correct solution. In other words, natural intelligence normally makes judgments based upon uncertain variables and unknown values, and the solutions are weighed in terms of success or failure of their consequences in some functional framework in the life-world of the organism doing the problem solving. So far, not machine has accomplished this kind of problem solving except in a very deceptively rudimentary manner. We can develop discrimination tables with predefined probabilities of outcomes and alternative uncertainty factors, but this kind of systematic approach to complex problem solving does not embrace the degree of intuition and experiential rationalization that is normally, automatically employed by natural creatures that are challenged in complex situations. It should be taken as a strength of natural human intelligence that though machines cannot solve even very simple natural problems, the human brain can rapidly and readily solve both mathematical, puzzle type problems as well as those problems requiring experience, intuition and application of reason and evaluation.

Related to the issue of abstract versus natural problem solving is the notion of learning and the quality of what is learned as a form of experience that can be applied to future problems. Again, machine learning has been advanced, but the parameters of such learning is almost always predefined such that learning cannot proceed beyond the purview of preprogrammed parameters. With natural intelligence, learning is almost innate and automatic to the context of application. The more intelligent the organism, the more it learns from the problems it encounters and solves. Another way of seeing this is that abstract intelligence is by definition programmable, or rather, pre-programmable and thus it is constrained by the parametric templates of its program. Naturally intelligent systems, beyond the long period of cultural-environmental acquisition and behavioral reinforcement that occurs on a continuous basis, are essentially non-programmable, or what we might say, it is innately or self-programmable.

Another means of describing these differences is to see the difference in functional outcomes of the two types of systems. Natural intelligence is extremely adaptable to functional variables in motor coordination. It takes a great deal of work to preprogram a machine to perform even simple tasks efficiently, much less complex series of tasks that natural intelligence seems capable of. On the other hand, abstract intelligence systems can perform complex linear calculations in a fraction of the time it requires naturally intelligent systems to perform similar kinds of abstract functions. We may say that natural systems are concrete and non-abstract, though they are capable of a limited degree of abstraction, as well as abstract intuition. Abstract systems tend to be non-concrete and incapable of extensive variable application, but capable of a rate and degree of abstract functioning that is preprogrammed.

I would venture the distinction between the two forms of intelligence as saying that abstract intelligent systems are primarily linear in function, whereas natural intelligent systems tend to be non-linear in function. Abstract systems are furthermore over-determined or at least fully determined systems, whereas natural systems tend to be critically underdetermined, hence normally chaotic systems.

I would be inclined to say that natural intelligence resembles more analog designs of intelligence, versus the digital form of intelligence that has arise with the machine information revolution. Digital intelligence seems to me to be fundamentally a form of abstract, mathematical intelligence, and hence suffers the limitations of design and possibility that are inherent to this form of information processing.