Abstract

A central issue in modelling, the "structural inference issue," is whether it is possible to ascertain, even allowing an infinite comparison of behavioral data, that a model truly represents the mechanisms underlying a real behavior. Employing the tools of the author's Theory of Modelling and Simulation (Wiley), a heirarchy of levels of system specification is introduced; the above question then becomes one of inferring from knowledge on one level structural information about an adjacent level--i.e., inferring that some sort of structure or function preserving morphisms holds between systems on different levels. Only under restricted circumstances, wherein certain justifying conditions are met, can one infer higher-level morphisms from lower-level ones. The paper explores some of the known justifying conditions and shows how the heirarchy enables consideration of the structural inference issue in a fundamental, rigorous manner. Epistemological implications of the analysis conclude the paper.