Introduction

If the sum total of all human knowledge were put into a machine and if that machine could make decisions based on that "knowing" it still could not match the human condition until it starts to learn from its own mistakes and, most essential, take pride in that learning which further engenders and motivates new learning. The keys here are motivation and emotion, things which we still hold to be uniquely human. We should keep this position clearly in mind as we progress towards the development of Metacognitive Cybernetic models of learning. Lars Qvortrup in CYBERNETICS & HUMAN KNOWING states:

While natural and social systems are autopoietical or second order cybernetic systems in that they produce not only their own elements and relations, but also their own conditionality, or - as Kant puts it - their own purpose, technological and semiological systems - artificial systems - are heteropoietical or first order cybernetic systems: they may or may not produce their elements and relations, but they are not conditionalised themselves. Finally, higher order artificial systems e.g. second order technological and semiological systems or "semi-autopoietical systems" (examples such as artificial intelligence and recursive pieces of music, respectively) are briefly characterized. While internally they produce their own condition in order for example to "mimic" a natural or a social system, externally they are conditionalised by the social system: their purpose is to mimic, it is not themselves. (Qvortrup, 1993)

We are clearly aware of our inability to create systems conscious of their own existence and able to change themselves as their environment changes (at least for the foreseeable future). Research arising from emerging technologies is bringing with it new demands for developing learning theories which seem to transfer into the realm of the machine that which we traditionally applied solely to human learning. Instructional technologists interested in applying existing learning theories to new models of instructional design utilizing these emerging technologies will find some theories don't quite fit, precisely because new capabilities exist where none existed previously. Adaptations and permutations of existing theory will evolve but we stand on the threshold of new learning paradigms as man and machine begin to truly collaborate in a common learning schema.

Some say we cannot hope to design such a system precisely because no learning theorists today can lay claim to truly understanding, much less fully modeling, how humans learn. If we can't design models which completely describe the processes of human learning how can we possibly design machines with such capabilities. Such a system would have to be metacognitively aware of all its components; capable of interacting with a learner; analyzing their myriad of needs; while growing, learning and adapting as the learner grows, learns and adapts. (It is a rare thing to find a human teacher capable of this.) Such a system does not seem remotely possible with todays technologies but there does exist elements within the fields of artificial intelligence and computer aided instruction which are beginning to move in this direction and can provide micro-blocks that deliver components of such a metacognitive cybernetic unit. Although we can't provide the whole model of human learning we have made amazing progress in describing parts of the process. What remains is to begin understanding the potential myriad of new technologies, combining them with proven learning theory and begin producing early MC models.

Instructional Technologists must be aware of emerging technologies as they begin designing their own learning systems and incorporate, where possible, such innovations as soon as it becomes feasible. It will soon become possible to have a program that can draw out of learners their own 'learning schemas' as they try to understand new concepts or processes and then begin developing learning modules to fit within that schema. As Osman & Hannafin state, "Although existing research and theory suggest that metacognition is integral to successful learning, existing instructional design (ID) models do not typically emphasize metacognitive strategies such as planning, monitoring, revising, and other self-regulating activities." (Osman and Hannafin, 1992) Such endeavors, by their very nature, involve a major rethinking of how we apply traditional learning theory to our instructional designs. "In attempting to simplify the learning in order to improve instructional efficiency and effectiveness, IST (Instructional Systems Technology) may be short-circuiting relevant mental processing. Designer's attempts to simplify learning risk supplanting the complexity that is inherent in the learning process or the task to be learned." (Jonassen, 1991) The clear point here is that the models we build must adapt, not just to the learning situation, but to the learner as well (an extremely daunting task). We should be able to take elements of existing learning theories and adapt components of them to the models we build; mixing behaviourism, cognitivism, constructivism and beyond where it proves successful for the learner. We should also be aware of how the learner, the teacher (the sometimes forgotten partner in this process), and the learning environments we design work together in a group dynamic. This step is essential as our learning environments become more sophisticated and take on roles which may rival that of the teacher.