General Systems Theory and Cybernetics:

The scientific disciplines of general systems and cybernetics have always been closely aligned, and often indistinguishable. But one of the clearest distinctions is in their relative strength of emphasis on philosophical integration (general systems theory) as opposed to the role of feedback/control mechanics (cybernetics). After all, the term cybernetics, which was coined by the distinguished mathematician, Norbert Wiener (1948), comes from the Greek word for "helmsman" or "steersman." Scientific contributors to both disciplines sprang from laboratories developing advances in communications and electronic engineering, mathematics, philosophy, physics, physiology, and even psychology, thereby making their collective research one of this century's most interdisciplinary of scientific efforts.

A relatively complete bibliography of the work in basic and applied general systems research published prior to the mid 1970's may be found in Klir and Rogers (1977). Some of the more notable and prolific contributors to general systems theory include von Bertalanffy, Rosen, Rosenbleuth and Wiener, and several other collaborative teams. Readers interested in the early articulations of cybernetics could begin with any one of a notable series of clearly written works by Ashby (1952, 1956), Walter (1953), Wiener (1954), Sluckin (1954), or Wisdom (1951). More technical contributions exist in Wiener (1948), Quastler (1953) and even in the technical series edited by von Foerster (1951).

In contrast to the sweeping theoretical and philosophical implications of general systems research, cybernetics placed a very strong emphasis on feedback as a self-regulating dynamic, and especially emphasized this phenomenon as it applied to machines, which researchers refer to as incorporating servomechanisms. Servomechanics is a phenomenon that was previously recognized in biological systems, and the Harvard physiologist, Walter Cannon, had already labeled the phenomenon, which he called homeostasis, well before the origins of cybernetics (Cannon, 1932).

Such distinctions of use and origins are especially significant for understanding the state of today's literature on adaptive instructional software systems. Servomechanical systems were originally designed to maintain functional outputs within some pre-specified range, with a mean within this range often being referred to as the "goal" of the system. The simplest example of such a "goal" oriented servomechanical system is an air-conditioned building. An air-conditioning system is installed to maintain a given air temperature within the building. The particular settings of a thermostat, usually under the control of the building's inhabitants, defines the desired temperature as the "goal state" of the larger architectural eco-system defined by incorporating the air conditioning system and the structural attributes of the building itself.

Thermostats have the capability of "sensing" the current air temperature, of "knowing" the desired temperature (setting), of "comparing" the difference between the two, and of "controlling" the activity of compressors/heaters to adjust the heating or cooling of the circulating air. Such a system may be defined as "homeostatic" in the same sense that Cannon used that term to describe physiological "steady-state" maintenance dynamics. That is, the building's rooms have a temperature that is "dynamically stable." By this we mean that the actual temperature is constantly varying (i.e., is dynamic--thereby representing the changing characteristic) slightly within some finite range, with that range's average being the thermostatic setting of desired temperature (which remains the same--thereby representing the stable characteristic of the system). Only extreme perturbations, will cause the room's temperature to vary outside of this narrow range. Such perturbations might include: (1) leaving a door or window open to allow outside air exchange, or (2) faulty insulation against normal interactions between internal and external temperatures.

The "feedback" in air-conditioning systems comes from the thermostat's ability to sense the constantly changing air temperature that it, itself, is controlling through its control of the compressor and circulation fans. When the thermostat senses that cooling has reached the desired set point ("goal"), it typically responds by turning the compressor and circulation fans off. (Actually it responds only when the air temperature is somewhat below the thermostat's set point, since systems are designed to respond to maximum/minimum tolerance range points, not the actual setting--which is what accounts for the ever-fluctuating operating characteristics that define the phenomenon of dynamic equilibrium.) When tolerance ranges are too broad within such systems, people are often seen re-setting the thermostat, typically accompanied by exclamations of, "it is too cold/hot in here!" In such cases, their individual sensitivity to maximum/minimum temperature fluctuations are more sensitive than the servomechanics of the thermostatic air controller.

Such feedback and control dynamics also define more advanced servomechanical devices, such as aircraft and missile auto-piloting devices. In all such systems, a goal is set and any major deviation of current activity away from that goal initiates action from the controlling mechanics of the system to return to that goal. Thus the operating characteristics of virtually all feedback systems incorporate oscillations of varying degrees, and are measurable in terms of both the amplitude and the frequency of those oscillations. When amplitudes are small, the system has fine tolerances. When frequencies are short, the system has highly responsive lag characteristics. And when long-term perturbations occur, such as breaking a window in an air-conditioned room, the system remains incapable of responding in any fashion other than trying to bring the state of things back to their set point.

This type of reaction to long-term perturbations has a lot to do with the ultimate failure of most human weight-loss programs. Humans have a "normal" body weight around which they fluctuate only slightly. In a weight-loss diet program, they may lose significant amounts of weight, but after their rather quick loss of weight, they find themselves gradually gaining all of it back within a few months. Their weight is homeostatically controlled by "set-point" types of dynamics within the brain structure called the hypothalamus. Severe and sudden disruptions in weight merely result in gradual recovery through the body's feedback and homeostatic control mechanisms.

Feedback and adaptive control within the context of instructional systems, such as MediaMatrix, relies upon sophisticated automatic knowledge generation engines which generate their representation of the student user's ever-developing knowledge via adaptive assessment and evaluation techniques. Such techniques assure that the most needed feedback has priority by selecting questions on content not yet assessed or content known to be weakly or wrongly learned to date. Such knowledge "mirroring" systems enable both adaptive guidance and adaptive tutoring services with the highest fidelity possible.

Next Topic... Adaptive Assessment