Close this search box.

Induction and Family Diagram

This article is an attempt to describe the essential concept of induction in science, how Bowen Theory needs truly inductive research, and how the motive behind the Family Diagram App and App Seminar is toward this problem.

While the Family Diagram app is used in a number of ways by different people, the driving motivation to develop the app has always been toward a more data driven science of human behavior. There are many difficult tasks ahead for a field of research Psychology that holds this aspiration. One preliminary task is developing a format to systematically organize and present concise theoretical arguments for legitimate scientific critique.


Induction is the act of superimposing an interpretation on a set of empirical observations. In other words, it is saying “Among these points, I perceive THIS pattern.” As a noun, an induction is the proposed pattern itself. Theories and concepts are inductions while algebraic equations and formal explicit models test them.

An induction is a perception to be tested. It makes a claim about how something works. It is not perfectly accurate, but a best guess.

An induction is tested by deducing an if-then prediction from it. The act of generating a hypothetical prediction, i.e. a hypothesis, from an induction is called deduction. The operation of induction is mostly intuitive while the operation of deduction is logical and mechanical.

Induction and deduction go hand in hand in the accumulation of knowledge. The researcher or R&D engineer alternates between the two as models for the problem at hand are incrementally refined. The key is to differentiate each operation from the other so that one knows which is occurring at any point in time.

This is important because regression in scientific thinking causes these operations to become less differentiated and more fused. It becomes harder to tell what is an assumption versus what is supported by data. Conclusions are then more deduced from existing models rather that updated through rigorous tests. Regression in groups result in over-agreement that reenforces existing models as dogma.

Consider an example of poorly differentiated induction and deduction: “We can see that the sun obviously comes up every day. Even when you can’t see it, it is there. We deduce all predictions from this truth.”

Now, the following is the systematic differentiation of induction and deduction on the same question regarding the behavior of the sun and earth.

You may observe that the sun comes up four days in a row. Based exclusively on those four data points, you may generate the induction: the sun comes up every day. You may then deduce a hypothesis from this induction that predicts the sun will come up every day over the next seven days. A formal experiment can then be run to see if the sun appears in the sky for each one of the next seven days. If the sun does appear in the sky each day, then the induction stands. In that case the induction is flawless but only so far as the collected data indicate.

The highest standard of rigor is to only claim what the data show, regardless of how intuitive or ridiculous this constraint feels. This keeps theories and models tied to facts alone. Anything yet unsupported by facts can be safely differentiated from inductions as speculation, and handled with the appropriate level of credibility.

Now assume that the eighth day is cloudy and the sun cannot be directly observed. How does one formally know that the sun is in the sky? Because the environment became more illuminated? What formal scientific models support that claim? Because the air became warmer? Again, what scientific models support that claim which supports the first claim that the sun comes up every day? This example is intentionally absurd so the it illustrates the seductive power of intuition versus the tedium of real scientific proof

Moving on, it is at this point in the process, where one feels the impulse to let a false prediction go, that induction comes back into the loop. Some creative explanation is required to get back to all claims supported by verifiable fact. In this case, the induction may be modified to “the sun comes up every day, where ‘comes up’ is defined as an increase in ambient lighting from ~0.001 lux to ~100,000 lux (Spitschan, Aguirre, Brainard, Sweeney, 2016), and an increase in ambient temperature of ~20 degrees Fahrenheit”.

The process can then continue with deduction to test this new induction, then induction to explain any and all new observed variation, then deduction to test, induction to explain, and so on. The models remain mechanical and verified. Unexplained variation, i.e. when the hypothetical prediction is not correct, must be accounted for for the theory to remain simultaneously based on fact alone and “solid.” The formal practice of measuring deviance from a perfect prediction is called “statistics.”

Where do new inductions come from? No one really knows. Where do any creative ideas come from? They certainly pertain to some logical structure but they are not generated purely by logic. Sometimes the best inductions appear in the mind on a walk, in a dream, or while bowling.

There is certainly no reasonable human on earth that spends time confirming that the sun reappears to the entire Earth every day without fail. Since the time of Copernicus, scientists have developed other inductions to explain the position of the sun in the sky in terms of their shape, mass, and relative positioning among other massive objects. The example here is intended to demonstrate rigor in differentiating assumption or speculation from fact. Being able to quickly identify the inductions and deductions at work in a particular statement about reality helps tremendously with this. This is the traditional practice of rational thinking that all courts, trade, politics, and the rest of the Post-Enlightenment Western is rooted in. Being unable to differentiate them leaves one vulnerable to regressing to assumption without evidence, and the darkness of blind dogma.

While it is as good as fact that the sun rises above the earth’s horizon every day, much of the scientific world remains fraught with methodological peril. Perhaps no scientific discipline is more perilous than the study of human behavior. In the world of psychotherapy it is still mainstream training to act on inductions that have never been made explicit, let alone tested. It is another practice entirely to lay out the logic behind one’s thinking and behavior in the form of a theory so that it can be formally critiqued.

The Status of Inductions in Bowen Theory

As of this writing, the many inductions of Bowen Theory do not enjoy the kind of systematically organized raw datasets required for critique from the broader scientific community. Further, experimental research that formally tests predictions from theory has never been published in respected, peer-review periodicals outside of the Bowen Network. Surely this must be a requirement for any “science.”

Part of this problem is technical; The “systems” inductions in Bowen Theory involve too many factors or variables to present in concise form for critics who are not primarily invested in digging deep into case studies. There are other factors, such as there being no well trained, professional scientific researchers that use Bowen Theory to guide their research in a substantial way.

The state of the art for arguing an induction in Bowen theory is the case study in literary publications and conferences. A proposition is made, for example that cutoff impacted one offspring’s ability to move smoothly through life more than a sibling. Then the case is presented over a good portion of the presentation. This long-form format requires an audience that shares enough of the bias of the presenter to devote substantial attention to a complicated heap of details. Further, the audience must imagine the difficult inductions without visual representation.

An important question is; What would it take to concisely argue an induction to a critical audience in three to five minutes? What would it take to do this for multiple cases in the same presentation or article?

The Family Diagram app aims to at least remove the technical barrier to organizing and presenting raw data on what theory calls “systems” phenomena. That would be the raw observations necessary to argue a particular induction pertaining to what theory calls an “emotional system” (Papero, 2015). To do this, the app was designed according to the essential characteristics of models of emotional systems (Stinson, 2020). This list is an induced average of the current verbal, behavioral, and written opinions of what Bowen Theory is today.

The Family Diagram app is just one way to organize the complex observations required to make the case that the family is an emotional unit. It is argued here that observations to test hypotheses from Bowen Theory are too complex to organize without specialized technology. Further, the Family Diagram app itself is not so much the focus here but the scientific pursuit that it enables.

Therefore, if a better tool or methodology emerges that enables family researchers to produce higher quality data sets for critique, then it deserves more attention than the Family Diagram app. Such an improvement would be critiqued in terms of the quality of the data it produces as opposed to speculation about how much better the tool might be. Until then any comparison would only be speculative.

There are plenty of areas of work to be done for Bowen Theory as a scientific theory. A list of example hypotheses is slowly being compiled here: Example Falsifiable Hypotheses for Research in Bowen Theory. This is just one effort. The Family Diagram App removing the technological barrier to data collection is another effort. The design for induction using the concepts of Bowen Theory, namely differentiation of self, is the main difference between Family Diagram and any other tools for diagraming genetic lineages or family systems, such as GenoPro, MS Viso, and others.

Family Diagram as an Inductive Tool

A first goal of the Family Diagram app is to remove the technological barrier to collecting and presenting a concise set of raw observations behind an induction. Traditionally, the only explicit framework for organizing such arguments is the static, two dimensional family diagram. A static family diagram does not communicate a dynamic, step by step emotional process. It also does not provide a necessary set of constraints to standardize a data set for a concise, three to five minuteargument.

Building on the rigor of the above example regarding the sun, this example illustrates rigor in presenting family system inductions to a critic:

Among all three of the Smith, Hansen, and Martinez families, I suspect I saw a similar pattern where tension rose between two people, and then the second one of them got together with a third person against the first. I am assuming this occurred from reading quotes X, Y, and Z from persons A, B, and C persons in each family. Then the first one, appearing agitated from having two against them,” tried to get back in with another against the remaining one. While motivations are inferred and not directly observed, I am assuming this motivation occurred from reading quotes X, Y, and Z from persons A, B, and C persons in each family. In the Smith and Martinez families, the pattern disappeared after things calmed down. I am assuming this occurred from reading quotes X, Y, and Z from persons A, B, and C persons in those two families. In the Hansen family, the “outsider” remained on the outside beyond the period of observation and later reported resentment. I am assuming this step occurred from reading quotes X, Y, and Z from persons A, B, and C in the Hansen family. I’ve decided to call this insider/outsider pattern a triangle. Three other families were also included in the experiment but showed increased tension but showed no such patterns. I am assuming this null result from reading quotes X, Y, and Z from persons A, B, C, D, E and F in those families. Further observations within each family are required to understand this variation in the proposed “triangle” pattern.

The above is all that is required to systematically present a data-driven induction. It is complete with an induction, the observations from which the induction was derived, and report of variation in the induction. Importantly, null results are included in the induction and no conclusion is drawn beyond what the data show on their own. The reader is then free to decide for themselves whether they agree that the observations, which in this case are direct quotes from the family members, should be properly coded “together with a third against the first,” or “outsider remained on the outside,” or as a null result.

Several studies with increasing sample sizes could then be conducted to replicate the experiment. Then all of these studies could be reviewed in a meta-study to determine the state of the research in that area. If pursued with rigor and written up in a professional manner, any one of these studies might be published outside the Bowen Network.

This sort of basic work has not been done for most ideas in Bowen Theory. This means that there is quite a lot of low-hanging fruit for family researchers. Even one of the simplest and most basic propositions in Bowen Theory, the triangle, would represent a substantial finding if demonstrated with data and correlated with some important problem like the modulation of anxiety in an individual.

However, it is also a possibility that such experiments are too inconclusive for mainstream publication. Or that it is not possible to obtain meaningful results within the attention span of the average hobbyist researcher. In this example there were three null results along with three positive results. Dealing with this variation raises all kinds of barriers to meaningful results if hypotheses are deduced straight from “theory.” Example problems are how to measure the level of anxiety/tension required for a triangle to occur, reliability of the definitions used for inside/outside positions, whether a long enough sample was taken from each family, etc.

Regardless of the myriad challenges for formal research under Bowen Theory, a first goal of the app is to finally make it possible to present a concise argument. Without the app, one has to figure out how to diagram the steps of emotional process with powerpoint slides or something similar. An audience also has to mentally visualize changes among multiple dimensions in a systems assessment over time. They must imagine this while also relating changes between them as well as between appropriate individuals in the family.

The app attempts to solve these technological problems by providing a family diagram that changes in accordance with a timeline. Conventional symbols of emotional process are added to the diagram and given dates and/or times to indicate when they occurred. Custom variables can be added as columns to the timeline table and visualized as the diagram iterates through time. All of this was designed to concisely present simple inductions derived from complex and disparate data.

Notably, all of this simplifies communicating what Bowen Theory describes as “emotional process” to people unfamiliar with the theory. Experience has shown that it is too difficult for a new audience to “hear” what emotional process means without a visual. While animated powerpoint is more useful than paper and imagination, specialized technology was required to produce the proper interactive visual.

Indeed, the inductions proposed by Bowen Theory are quite simple. For example, anxiety impacted symptom in person C, family impacted anxiety via a pathway from person A, then B, then C. The problem has always been organizing the myriad observations into a view that is concise enough to argue those simple inductions.

The work now is to organize raw observations from real cases into the app. This remains a task for many.

The Role of the App Seminar

As stated above, many people use Family Diagram in many different ways. Broadly speaking, there is a spectrum of use from using the diagram as more of a drawing tool to systematically challenge existing assumptions through data. The goal of this article is to differentiate the latter, inductive mode of use from the former, deductive mode of use.

An argument could be made that it is the aspiration toward data-driven scientific research that differentiates Bowen Theory from many other research and professional traditions in Psychology. As discussed, there is still much work to be done for this aspiration to become a reality.

The current motivation behind the App Seminar is to promote and develop that inductive mode of use. Because the app is a tool, it has a strong gravitational force toward how-to questions of technique that kill systems thinking in a seminar. Therefore, the seminar was framed around falsifiable thinking. So far this frame has proved an equalizer for discussions in the seminar. Anyone can bring any wild hunch or abstract speculation to the table because everyone equally suffers the burden of proof.

The app makes conversations about proof explicit and practical instead of abstract and speculative.

As of this writing, the App Seminar is meant to set the bar for the inductive mode. This is evident in the fundamental requirement for entry, the ability to define a falsifiable prediction based on Bowen Theory in a manner suitable for data collection. In other words, the requirement is to meet the standards of mainstream science and medicine, regardless of how simple the prediction, experiment, or project.

That point is worthy of repetition: regardless of how simple the prediction, experiment, or project.

Thus, the two key components of the seminar are falsifiable thinking and project suitable for data collection using the app. The app’s requirements for data collection are no different than the requirements for Bowen Theory itself, except that the app requires one to do it explicitly and systematically. While these two components may or may not be ideal, they function to encourage systems thinking within concrete reality.

This is where the first aim of the app comes to some fruition: The burden of collecting data into a predefined format quickly exposes unclarity in thinking, from vague definitions to cherry picking data. It is terribly difficult to prevent such regression to undifferentiated induction and deduction. The explicit format of the app makes it possible for others to critique one’s assumptions, definitions, and inductions in concrete terms.

Most interestingly, very few have achieved a cogent case formulation with the app, for example even with as few as 4-5 observations. The best example thus far is Dr. Laura Havstad’s “Blue” case example. It is highly recommended to view Dr. Havstad’s example.

Most folks who seriously pursue the inductive mode discover that their data was not yet organized or not yet complete where they previously believed it to be of high quality. Most interestingly, this occurs in almost every case. Thus, the app itself appears to have the effect of demonstrating the real work to be done.

Once this is discovered, organizing observations can just as easily happen in excel or on paper. In fact, this step doesn’t have to have anything to do with the app. But the goal of eventually entering the observations into the app forces thinking and work to become more clear.

Thus, the current aim of the App Seminar is to hold focus on factors that increase this effect. Whether within or without the app, work in the seminar continues in the frame of falsifiable observations and predictions.


Papero, D. V. (2015). The family emotional system. In R. J. Noone & D. V. Papero (Eds.), The family emotional system: An integrative concept for theory, science, and practice (pp. 15–28). New York, NY: Lexington Books.

Spitschan, M., Aguirre, G., Brainard, D. H., & Sweeney, A. (2016) Variation of outdoor illumination as a function of solar elevation and light pollution. Nature, Sci Rep 6, 26756.

Stinson, P. (2016). To What Extent did the Buddha Define a Natural System?: Insights from Bowen’s Natural System Theory of the Family. (Doctoral dissertation, California Institute of Integral Studies). Retrieved from

Sign Up For News

Stay current.