Search
Close this search box.

Some Guidelines for Natural Systems Models

The following is taken from my 2020 dissertation which attempted to clarify the paradigm of emotional systems that Murray Bowen’s theory of the family existed in. The ideas have been updated somewhat since this writing, and I have published a video outlining the newest version. Nevertheless, this written chapter is the most verbose and detailed description of these ideas, which aim to define the basic principles a theory must follow before it can be compared with Bowen Theory. If there is a mismatch with these principles, the theory consists of a different project entirely and is incommensurate with Bowen Theory.

The BT literature suggests that the theory describes a biological phenomenon through a unique view not yet precisely replicated in biology. This unique view is a natural system theory and is a product of a kind of methodology particular to the development of such a theory. It appears as though no formal definition of such a theory and its respective methodology exists in the literature. Therefore, a preliminary goal of this study is to infer the guidelines for both the theory and methodology from an exhaustive review of the Bowen literature.

The goal of a natural system theory is simply to achieve greater validity than is possible with a linear–reductionistic theory. Such conventional models are less capable of accurately accounting for phenomena as they become more complex. A natural systems model may augment a conventional model using these guidelines. Each guideline represents a metatheoretical dimension which can determine the strength of a systems methodology. As in a conceptual system theory itself, each dimension describes a distinct aspect of such a theory but is also dependent on the others.

Account for all recorded observations

Perhaps the most unique guideline for a natural systems methodology is the effort to produce a model which accounts for all recorded observations. This implies a greater degree of rigor and careful thinking that ensures every change to the model accounts for all recorded observations. Popper (1963/2002) argued that Freud and Adler had only proven that they had created theories that could explain all observations but had not proven that their explanations were valid. Therefore, the unique challenge of a natural system theory is perhaps to retain the ability to account for all observations while also retaining the criterion of falsifiable predictions.

This characteristic is covered in a single quote from Bowen’s (1978) book, “When research observations were not consistent with the hypothesis, the hypothesis was modified to fit the new facts, the psychotherapy was modified to fit the hypothesis, and new predictions were made about the results of the psychotherapy” (p. 470). The key phrase in this passage is “the hypothesis was modified to fit the new facts.” While this might appear a self-evident principle of every scientific methodology, this step is rarely, if ever, accomplished in nonreductionistic research. The hypothetical result is a perfectly predictive theoretical model that transcends the function of statistics. This is because descriptive statistics measure the degree to which a phenomenon adheres to an assumption generated by a human. The kind of natural systems model suggested here simply produces factual predictions about how processes play out under specified conditions.

The most challenging aspect of this rather idealistic guideline is only implied in this quote. This is the step of refining the rich concepts in the model so that they simultaneously increase in breadth and precision. This cognitive operation is called induction in traditional logic, which describes the creative production of general law from observational fact. Interestingly, the term has also adopted from logic to describe the process through which one group of cells influences another to assume novel functionality in complex systems biology (Kaneko, 2006). In experimental research aimed at increasing theoretical validity, the induction phase works in tandem with the deduction of hypothetical predictions from the model. A new concept or variable is induced and then tested by deducing a prediction from the model and comparing that prediction with observations. When the newly induced variable or variables do not accurately account for every recorded observation, the induction operation must be reexecuted until the model matches the observational facts. The induction operation is literally creative because it “intro-duces” a new idea into the model while the deduction operation is simply a logical derivation of the model (Kaneko, 2006). In other words, two people may induce different explanations for a single set of facts, but two people should deduce a single prediction from a single model. It is the vulnerability imposed by the experimental process which determines which induction is less valid than the other.

The move toward analysis in conventional research sacrifices long-term comprehension of the total phenomenon for initial precision by reducing the number of variables involved. Such a reductionistic model shifts to measuring how much the data adheres to a human assumption, for example a central tendency in statistics, rather than simply predicting the behavior of the system. The sheer difficulty of the inductive operation itself may account for the overwhelming predominance of conventional, reductionistic models in science. Such models bias toward the precision of analysis and away from the breadth of synthesis. Perhaps the kind of systems model described here requires an integration between analysis and synthesis that is simply beyond the evolutionary sophistication of the human species at this point in time.

Describe processes, not simply content

Perhaps the simplest definition of natural systems thinking is the observation and description of processes of nature as opposed to simply the content of the processes of nature. This simple characteristic can immediately increase the level of objectivity in a model by producing falsifiable predictions based on observations as opposed to simply producing abstract ideas related to human assumptions.

The word system is often used in many different ways, for example, to describe a categorical model as a system with no description of how it is a system (M. Kerr & Bowen, 1988). One such example is Bronfenbrenner’s ecological system theory, which describes the categories of variables implicated in human development but does not point to the verifiable processes that account for the categories and their relation to one another (Bronfenbrenner, 1979). Categorical models such as Bronfenbrenner’s or the DSM nosology (American Psychiatric Association, 2013) may encourage linear–causal, reductionistic thinking. Such categorical thinking may reduce the complexity of the phenomenon in question around a human assumption rather than the observational facts. This type of linear–causal thinking is evident in an overemphasis on efficacy studies and underemphasis on theoretical research, where the former represents something more similar to an engineering discipline and the latter represents something more similar to a scientific discipline.

Consider concurrent reciprocation

The interaction of multiple, concurrent processes implies the emergence of chaotic behavior in a complex system. Gleick (2011) describes the qualitative shift that occurs when accounting for the concurrent interaction of three variables versus only two. Perhaps one example of this problem would be the difference between the interaction of three gravitational fields versus two gravitational fields. In BT an example would be modeling the emotional transactions between a hypothetical two-person system and a hypothetical three-person system. Such a chaotic system could be described as exhibiting concurrent reciprocation, or the interaction of multiple feedback loops which operate simultaneously through time.

Describe a change in one variable in terms of a change in other variables

In “A Systems Model for Disease,” Michael Kerr (1980) describes a systems model as one which defines how an equilibrium is maintained between a set of variables. That is, if a system is described by five variables, then the model would predict the values of the remaining four variables in response to a change in one variable. This model implies that the variables represent dimensional continuua as ordinal, interval, or ratio values.

A hypothetical example from my own understanding of M. Kerr’s (1980) writing may illustrate the characteristics of a systems model. For example, an ideal systems model could normalize the raw values of each variable to fit in a range of 0.0–1.0 and require that the sum of the normalized values of each variable always equals 1.0. The choice of 0.0 and 1.0 is arbitrary, as any two constants would do. Assuming a system consisting of five variables, equilibrium would be represented by the sum of all variables in a set equaling 1.0, as in the set {.2, .2, .2, .2, .2}. If one variable were shifted up, as in the set {.3, .2, .2, .2, .2}, then the system would be out of equilibrium because the sum of the set is now 1.1, which is greater than 1.0. Any or all of the remaining four variables would need to change to compensate for the positive change in the one. The model would determine how the load of compensation would be spread among the remaining four variables.

There are an infinite number of combinations which would bring the system back into equilibrium in this idealistic mathematical model. One example is {.3, .2, .2, .2, .1}, where only one variable compensated with a negative shift. Another example is {.3, .2, .2, .15, .15}, where two variables compensate evenly with negative shifts. Another is {.3, .2, .1, .4, .0}, where three variables compensate. However, in this example two variables compensated negatively to varying degrees and one actually compensated positively and with a shift greater than the shift in the original variable. These permutations demonstrate a concrete distinction of a systems model, but also the complexity inherent in reciprocal interaction between multiple dimensions. Though technically deterministic, this behavioral complexity is sometimes characterized as “chaotic” (Gleick, 2011, p. 39). A systems model itself may be defined as the set of rules that determine the distribution of each compensation. While the mathematical system model above might accurately describe a simple mechanical system, each variable in a living system is probably changing simultaneously. Therefore, the project of modeling a living system is likely far more complex than the ideal example given here. BT suggests that the reciprocal functioning of each member of a family can be modeled more or less in the manner described above. In fact, Bowen claims to have developed such a model of the family and used this model to generate the predictions used in his initial experimental research at NIMH. Unfortunately, the details of his model do not appear to be available in the literature. Because the functional level of each member of a family is continuously managing a balance of functioning for self or for the system, their individual functional level is in constant motion. The guidelines for a natural system theory reviewed here are simply suggested to increase the precision of such a dimensional model.

Further, more complex systems may describe hierarchies of variables as opposed to placing each variable at peer level. For example, in a family there may be multiple variables to consider in determining the portion of each individuals functional level, which are determined by the individuality force present within them and not the togetherness force in the family. Capturing such complexity may initially be a daunting task. Nevertheless, the takeaway from this guideline is simply that a systems model defines how one variable or set of variables changes in response to a change in another variable or set of variables, often in service of maintaining an equilibrium.

Relate all variables to every other variable

A systems model might avoid the use of discrepant models to account for the phenomenon. Categorical models tend to reinforce the notion of isolated phenomena by imposing value-laden notions of normal and abnormal states. A complete system theory might seek to relate all aspects of the system. This concept is in line with the European Enlightenment vision of a unity of knowledge (E. O. Wilson, 1998) and is perhaps more critical in a natural system model and implies a greater degree of internal consistency.

This characteristic can be expanded to include relating theoretical conceptualizations to the rest of the sciences. At the very least, a natural systems research methodology must be available for methodological critique by adjacent scientific disciplines. Counter examples to this are N-of-1 psychological and sociological methodologies which assume that immediate subjective experience is the most valid data for experimental research (Ponterotto, 2005). Such methodologies may rely on a first-person account of nonrandomized samples (Baldwin, 2018; Smith, 2018; H. Wilson, 2017) as opposed to a randomized trial with sample sizes with optimal statistical power. The implicit assumption is that human experience occurs in a compartment apart from the sciences. While a first person account may inform exploratory experimental research, this class of methodologies is not accepted by the broader range of scientific disciplines whose exclusive aim is the development of valid and reliable theory, such as physics, astronomy, chemistry, meteorology, and so forth (Kazdin, 2016). In the context of natural science, theories developed using these methodologies can be used for hypothetical speculation but cannot enjoy the status of scientific research or accurately inform public policy without methodological peer review by adjacent disciplines in the natural sciences (E. O. Wilson, 1998).

Generate falsifiable predictions

Popper’s (2002) criterion of falsification ties theory in the objective domain. If a weakness of linear reductionistic thinking is the assumption that understanding a part is adequate to understand the whole, then a weakness of holistic thinking may be the assumption that identifying the whole implies an understanding of the parts. Popper’s criterion provides a counterbalance to this weakness in holistic thinking as a reinforcement of dogmatic belief. A falsifiable prediction is one where its validity is self-evident by virtue of its accounting for all variables implicated in the prediction.

The current gold-standard in psychological research is the double-blind randomized control trial which is automatically organized around falsifiable prediction (Kazdin, 2016). Blind predictions can be produced by an accurate systems model, where a change in one variable is accounted for by changes in one or more other variables. Bowen claimed that his model of the family system predicted every outcome and accounted for every observation. Unfortunately, it is unclear from the literature how subjective these predictions were. It is nevertheless compelling to review the degree of objectivity in the class of predictions he claimed to have made. In the original NIMH project (Bowen, 2015), he claimed that a positive change in the mother was followed in a number of hours by a negative change in the daughter, and vice versa. He reported that these blind predictions were so accurate that the staff would prepare for the onset of the symptoms in the one person after observing the remission in the other person (Bowen, 1978). The false outcome of this prediction would be if the staff prepared for an onset or remission of symptoms that never occurred. A falsifiable experiment simply asks: Did the predicted outcome occur or not?

One important fallacy of systems thinking may be what Lilienfeld et al. (2015) termed the mantra of holism. The mantra of holism is one way that holistic or systems thinkers may explain away critiques of the validity of their model by stating that theoretical conclusions cannot be critiqued outside a broader context. This claim in itself is not problematic, but it becomes a methodological fallacy once the researchers also fail to accurately account for that broader context. A systems model which declares reciprocal interdependence between its own variables must take this into account. The fallacy is avoided by producing falsifiable predictions which account for all variables implicated in the outcome.

Use terms independent of the existence of Homo sapiens

It is a choice to describe human behavior as a distinctly human system or as a natural system. The former may emphasize the differences between humans and the rest of life and reinforce subjective impressions; the latter may emphasize the similarities between humans and the rest of life and challenge subjective impressions. If a systems model describes human behavior as the functioning of natural systems, then it can move more rapidly toward validity by using terms that do not depend on the existence of humans. This guideline is predicated on the assumption that the human species is an outgrowth of deeper natural laws common to all species as well as inanimate matter. It implies that the difference between human behavior, ant behavior, and ultimately the behavior of subatomic particles is proximally more quantitative than qualitative in the experimental process but distally quantitative and not qualitative. Counter examples to this guideline are the use of terms derived from human history and culture to describe instinctual behavior which has existed far longer than human history and culture. This may even include the terms objectivity and subjectivity themselves, as they are not equally used to describe the behavior of animals, insects, and plants.

This guideline implies the possibility of an organismic theory. Such a theory would relate what are now called biological phenomena to basic laws of physics and chemistry (Elsasser, 1987/1998). It would make descriptions of the behavior of any species exist in terms not dependent on the existence of that species but merely dependent on existence of the material aggregates that compose it. At this hypothetical point in time, what is now considered the domain of biology would simply describe one degree of complexity in the organization of materials.

References

Baldwin, Richard. (2018). Who are they to judge? Overcoming anthropometry and a framework for fat bodybuilding. Fat Studies: An Interdisciplinary Journal of Body Weight and Society, 7(3), i–xiii. doi:10.1080/21604851.2018.1453622 (Retraction published 2018, Fat Studies: An Interdisciplinary Journal of Body Weight and Society)

Bowen, M. (1978). Family therapy in clinical practice. New York, NY: Jason Aronson.

Bowen, M. (2015). The origins of family psychotherapy: The NIMH family study project (J. Butler, Ed.). Lanham, Maryland: Rowman & Littlefield.

Bronfenbrenner, U. (1979). The ecology of human development. Cambridge, MA: Harvard University Press.

Elsasser, W. (1998). Reflections on a theory of organisms: Holism in Biology. Baltimore, MD: The Johns Hopkins University Press. (Original work published 1987)

Gleick, J. (2011). Chaos: Making a new science. New York, NY: Open Road Integrated Media.

Kaneko, K. (2006). Life: An introduction to complex systems biology. New York, NY: Springer.

Kazdin, A. E. (2016). Research design in clinical psychology (5th ed.). Boston, MA: Allyn and Bacon.

Kerr, M. (1980). A systems model for disease. In R. R. Sagar (Ed.), Bowen Theory and Practice: Feature articles from the family center report, 1979–1996 (pp. 133–137). Washington, DC: Georgetown Family Center.

Kerr, M., & Bowen, M. (1988). Family evaluation: The role of the family as an emotional unit that governs individual behavior and development. New York, NY: W. W. Norton.

Lilienfeld, S. O., Lynn, S. J., & Lohr, J. M. (2015). Science and pseudoscience in clinical psychology. New York, NY: The Guilford Press.

Ponterotto, J. G. (2005). Qualitative research in counseling psychology: A primer on research paradigms and philosophy of science. Journal of Counseling Psychology, 52(2), 126–136.

Popper, K. (2002). Conjectures and refutations: The growth of scientific knowledge. New York, NY: Routledge. (Original work published 1963)

Smith, M. (2018). Going in through the back door: Challenging straight male homohysteria and transphobia through receptive penetrative sex toy use. Sexuality & Culture, 22(4), 1542–1542. doi:10.1007/s12119-018-9536-0 (Retraction published 2018, Sexuality & Culture)

Wilson, E. O. (1998). Consilience: The unity of knowledge. New York, NY: Alfred A. Knopf.

Wilson, H. (2017). Human reactions to rape culture and queer performativity at urban dog parks in Portland, Oregon. Gender, Place, & Culture: A Journal of Feminist Geography, 1–20. doi:10.1080/0966369X.2018.1475346 (Retraction published 2018, Gender, Place, & Culture: A Journal of Feminist Geography)

Sign Up For News

Stay current.