Based on the provided sources, the observer decides which variables define the system through a process of selection driven by specific goals and the necessity of finding a predictive model. Ashby emphasizes that a “system” is not a physical object, but a list of variables chosen by the observer from the infinite number of variables available in any material thing[1].
Here is how an observer decides which variables to include:
1. The “Generating Question” and Main InterestThe primary criterion for selecting variables is the observer’s specific purpose or “main interest”[2][4].
• Pragmatic Selection: Selection cannot be purely arbitrary; it must be guided by a motive, such as curing a disease or fixing a machine[5]. A “generating question” (e.g., “How do I reduce delinquency in this town?”) provides the criterion for which variables are relevant[4][6].
• Relativity: Different observers will define different systems on the same object based on their interests. For example, a neurophysiologist defines a sheep’s brain as a complex system of enzymes and fibers, whereas a butcher defines the same object as a simple system distinguished from about thirty other “meats”[7].
2. The Criterion of State-DeterminatenessOnce a provisional set of variables is chosen, the observer applies a rigorous test: is the system state-determined? This means, does the current state of the selected variables uniquely determine their next state?[8][9].
• Trial and Error: If the selected variables behave “capriciously” or unpredictably, the observer assumes an active and relevant variable has been left out[9][10].
• Refining the List: The observer adds or removes variables until a set is found that yields a single-valued, predictable transformation[10].
• Example: If an observer studies a pendulum using only the variable “angular deviation,” the system will not be predictive (it is not state-determined). By adding a second variable, “angular velocity” (or momentum), the system becomes state-determined and predictive[11][12].
3. Testing for Relevance and ConditionalityThe observer decides to include a variable if it is “relevant,” meaning it has a demonstrable effect on the other variables in the system.
• Operational Test: The test for relevance is operational: does a change in variable Y cause a change in variable X? If so, X’s behavior is “conditional on” Y, and Y should be included[13][14].
• Independence: If a variable is found to be independent (its variation has no effect on the others), it may be discarded or treated as a separate system[15][16].
• Completeness: The goal is to find a set that is “closed” or complete, meaning that for every variable in the set, all variables that affect it are also included in the set[13][17].
4. Handling “Memory” and Unobservable VariablesIf a system is not predictable based on the observable variables, the observer is often forced to invoke the concept of “memory”[18].
• Coding the Past: “Memory” is not an objective property but a reflection of the observer’s inability to see all necessary variables[19][20].
• Restoring Predictability: To make the system determinate, the observer must either discover the hidden variable causing the behavior or include the system’s past history as a variable in the definition[21][22].
5. Nature of the VariablesAshby notes that the observer is not limited to numerical or metric variables.
• Qualitative Variables: The observer may select nominal variables (e.g., “blood type,” “cloud type,” or “switch on/off”)[23]. The only requirement is that the variables can be clearly distinguished and observed[26].
• Time: While time is recorded, it is generally not included in the list of variables that define the system; rather, it is treated as the framework within which the system changes[27].
In summary, the observer defines the system by selecting a set of variables that is relevant to their specific goal and sufficiently complete to allow for determinate prediction of future states[2].
References
[1] Adaptive Systems work of ashby and ultrastability version 2.pdf [2] Ashby Mechanisms.pdf [4] Ashby Mechanisms.pdf [5] Ashby Mechanisms.pdf [6] Ashby-Mechanisms_of_intelligence_ocr.pdf [7] Klir - Ashby Foreword.pdf [8] Adaptive Systems work of ashby and ultrastability version 2.pdf [9] [Book] Ashby - Design for a Brain The origin of adaptive behaviour.pdf [10] [Book] Ashby - An Introduction to Cybernetics.pdf [11] [Book] Ashby - An Introduction to Cybernetics.pdf [12] [Book] Ashby - Introduction to Cybernetics.pdf [13] Ashby Mechanisms.pdf [14] Ashby-Mechanisms_of_intelligence_ocr.pdf [15] Ashby Mechanisms.pdf [16] Ashby-Mechanisms_of_intelligence_ocr.pdf [17] Ashby-Mechanisms_of_intelligence_ocr.pdf [18] [Book] Ashby - An Introduction to Cybernetics.pdf [19] Ashby 1991 - Facets of Systems Science.pdf [20] [Book] Ashby - An Introduction to Cybernetics.pdf [21] Ashby Mechanisms.pdf [22] [Book] Ashby - Introduction to Cybernetics.pdf [23] Ashby Mechanisms.pdf [26] [Book] Ashby - Design for a Brain The origin of adaptive behaviour.pdf [27] [Book] Ashby - Design for a Brain The origin of adaptive behaviour.pdf