Towards a system-theoretical decision logic

Abstract

A system-based decision logic predicated on subjective and objective probabilities is developed incorporating the Bayesian learning process. Selection of specific analytical instruments for generating informational stock pertaining to the system under investigation is described by learning curves which empirically treat either growth in raw information stocks or- which, as a corollary, empirically measure the reduction in expected error associated with models of system phenomena. The decision logic is extended for handling shifts in instrumental modalities, that is, switching from one instrumental category to another during the analysis process. Thus, selection of analytical instruments, and development of system-analysis strategies, need not be totally a prioristic. Although the procedural paradigm presented here is still somewhat immature, it may help focus attention on opportunities for ‘optimizing’ analytical and system modelling processes.