Behavioral systems are those whose mechanism cannot be observed or studied directly. The reasons they cannot be so studied may be that they are too complex—comprise numerous elements that interact in complicated incomprehensible, intractable ways, or their insides are not accessible. Hence such systems must be treated like black boxes. Examples of behavioral systems include: human beings, human intelligent or creative behavior, societies, armies and other organizations, mice behavior in a maze, and possibly even some very complex computer programs.

A strategy to investigate behavioral systems is to collect very large data sets that record inputs (stimuli—whether circumstantial or administered) and outputs. Typically these data refer to a large sample from a population of like-systems, but designs that focus on a small sample or a single system at diverse states are also possible.

This strategy requires appropriate methodology and techniques. Grossly speaking the challenge here is twofold. The first concerns the analysis of variables (stimuli presented-- by design or by Nature—to the investigated systems, and responses obtained therefrom) that, on the one hand, are very numerous, and, on the other hand, form but a sample from of the set of all conceivable variables; that set being typically infinite. The second concerns the identification of latent—unobservable—variables that in fact govern the behavior of the system.

1/ Analysis of a large sample of observed variables. Typically in statistical thought, both the increase in the number of variables, and possible interdependencies among them present a challenge to be dealt with, a burden to be surmounted. Techniques like regression analysis are offered to meet this challenge with varying degrees of success. The variables to be observed are selected for being of interest in their own right.

In contrast, in Facet Systemic Modeling (FSM), the approach to be presented here, the variables observable-in-principle (in effect, all behaviors) are assumed to be naturally interrelated –by virtue of pertaining to the same system-- and, moreover, their number is assumed to be infinite. The scientific imagery adopted for all possible system behaviors is that of a spatial continuum, a topological manifold embedded in a geometrical space of suitable dimensionality. Behaviors observed in a particular study are regarded as but a sample from all possible behavior, and are of no special interest in themselves, but rather as clues from which inferences are to be made concerning the totality of possible behaviors. (Thus, it may be said that in FTM the perspective on the variables is top-down rather than bottom-up.) The data analyses are carried out by Faceted Smallest Space Analysis (FSSA, Shye, 1991; 1999; Shye & Elizur,1994; Borg & Shye, 1995), a modern development of the MDS family of computer programs that incorporates substantive systemic considerations, and by the Partial Order Scalogram Analysis computer program (POSAC. Shye, 1985, 1999; 2008).

2/ Inferring Nature's underlying parameters. The search for such parameters is a major purpose of the science endeavor in any field of investigation. The quest is for relatively few underlying variables that determine the multitude of observed phenomena. The present author has shown that, in contrast with other multivariate theories and techniques, there are in fact not one, but two essentially different kinds of such parameters (fundamental variables). The one kind (facets) is for structuring the system behaviors, the other kind (coordinate-scales) for assessing qualities for the systems themselves. The former can be discovered by faceted data analysis (thru Faceted SSA applied to the behavior concept space), the latter by partial-order analysis (thru POSAC applied to the set of observed behavior combinations (systemic states)). Mathematical relationships between the two spaces, the behavior concept space and the measurement space have been investigated and interesting initial results obtained.

In this paper I propose to show how elementary considerations and simple axioms enable a classification of system-relevant events according to their significance to the system. From which a structure of systemic functioning emerges, and clear testable hypotheses follow concerning the structure of the system. Next, partial order dimensionality will be defined and measurement space for systemic states, defined in terms of systemic functioning, will be presented. The concepts and procedure will be demonstrated on two specific complex systems: The system of human individual functioning (defined as systemic quality of life, SQOL). Based on previous work (Shye, 1989), it will be shown that application of the faceted systemic modeling to human QOL results in four human functioning modes interpretable as the cultural, personality, social and physical modes. Viewed as sub-systems in their own right, these modes can be further analyzed, and through successive nesting operations finer modes can be identified. (By the way, the systemic quality of life model (Shye, 1989) was rated first among 68 models published in the 37 years 1965-2001. See Taillfer et al. 2003). New applications to real quality of life data in social (immigration) and in the psychological (psychotherapy) contexts will be presented.

The system of intelligent/creative behavior. In contrast with approaches that study human cognitive behavior by attempting to follow the mechanism of intelligent creative thinking in problem solving, the behavioral-systemic approach is content with regarding the human brain as a black box, studying its functioning through its manifest behavior, as in the common psychometric testing. Again, the power of this approach is in its capability to deal with large numbers of systems and of observed variables, and treat both kinds of entities (the systems as well as the variables describing their behaviors) as samples from larger universes, identified as The System and the Behavior in question (Intelligence etc.). New results will be presented from intelligence and creativity research, that (a) structure and inter-relate the two, (b) suggest meaningful justifiable scales for their measurement.