Wallis, S. E. (2016). The science of conceptual systems: A progress report. Foundations of Science, 21(4), 579–602.
I wish to express my appreciation for the reviewers’ close reading and detailed suggestions which have resulted in a more readable and effective paper. All remaining mistakes and weaknesses belong to me.
In this paper I provide a brief history of the emerging science of conceptual systems, explain some methodologies, their sources of data, and the understandings that they have generated. I will also provide suggestions for extending the science-based research in a variety of directions. Essentially, I am opening a conversation that asks how this line of research might be extended to gain new insights—and eventually develop more useful and generally accepted methods for creating and evaluating theory. This effort will support our ability to generate theory that is more effective in practical application as well as accelerating the development of theory to support advances in other sciences.
This paper reports on the “science of conceptual systems.” Metaphorically, each conceptual system provides a lens or framework of understanding which enables the user to understand and engage the world. Generally, a conceptual system is any form of theory, model, mental model, policy, etc. Here, because we are addressing a wide range of disciplines—and each discipline has its own language—we will use those terms interchangeably.
The purpose or direction of this science is to gain a better understanding of conceptual systems so that we can create better ones. Better conceptual systems are those that allow us to more effectively understand and engage the world—to reach our goals more easily. Here, conceptual systems are our primary source of “data.” And, our methodologies are focused on the relationships between those data. Additionally, we compare the usefulness of the conceptual system with their effectiveness when applied in the world.
As we all know, the world faces grave problems—including poverty, war, injustice, crime, and economic instability. To understand and resolve these kinds of problems, we turn to science. Because these are generally social problems, we turn more specifically, to the social/behavioral sciences.
A science may be defined as “the pursuit of knowledge and understanding of the natural and social world following a systematic methodology based on evidence” (Science Council, 2010). Science includes measurement, data, experimentation, observation, critical analysis, and other closely related requirements. A science of conceptual systems may then be understood as the pursuit of knowledge and understanding of conceptual systems using rigorous methodologies.
Since the scientific revolution, science has been generally successful in the process of finding facts. Using inductive, deductive, and abductive approaches, science has developed new ideas and insights, used data to build theories, and tested theories to develop more data. Within the natural sciences, these kinds of efforts have led to the development of very useful theories and laws. Within the social/behavioral sciences, however, such efforts have not been so successful in creating more effective theories.
While most of the social/behavioral sciences have been focused on empirical studies, there seems to be an underlying assumption that once we have enough data (or the right kind of data, or data from well-funded research) that useful theories will somehow emerge. This akin to believing that if we have enough bricks, a house will somehow build itself. In short, the “normal” view of theory building has not proved effective in advancing the social/behavioral sciences. While there have been inroads, those paths seem to have led mainly to cul-de-sacs of science.
One direction of thinking suggests that each theory is a lens or frame that we use to understand and/or engage the world around us (Bolman & Deal, 1991. For example, one may look at a business as a supportive family or a political battlefield—and take very different actions, accordingly! It is undoubtedly useful to be aware of one’s frames of reference, particularly in a business setting (Senge, Kleiner, Roberts, Ross, & Smith, 1994). However, those kinds of approaches have generally been fuzzy. They have not quantified their findings. Thus, the ability to create more effective theories has been elusive. Without rigorous methodologies, it is difficult to think of such efforts as fully scientific. Indeed, in trying to describe a good test for business strategies, Parnell (2008) could find no reliable advice. Instead, he suggested that organizational leaders rely on “artistry.” Similarly, others rely on intuition to evaluate theories (Shotter & Tsoukas, 2007). Like artistry, however, intuition is not a reliable tool (Meehl, 1992: 370).
Another direction suggests that it is not possible (or even desirable) to create formal theories (Burrell, 1997; Shotter, 1994). There, it seems that the authors are trying to “prove a negative”—a very difficult task. And, in that process, they implicitly create their own theories. Therefore, I am inclined to reject their claims as tautological.
Part of the classically accepted path is that theory may be evaluated by testing its generalizability, sensemaking ability, testability, plausibility, and usability (Sussman & Sussman, 2001: 90-92). Kaplan (1964: 311-322) suggests tests based on norms of correspondence (does the theory fit observed facts), coherence (does the new theory fit the existing theory), and pragmatism (does the new theory work when it is applied). Correspondence to “facts,” however, is problematic because all observations are theory-laden. So what counts as a fact depends very much on the theory. Also, what is pragmatic for one individual may be abhorrent to another. Finally, if a new theory coheres (matches) perfectly to the existing theory, they would be identical—and there would be no new theory.
Within the social sciences, it is often difficult to evaluate the application of a theory. For example, in the sphere of economic policy, we cannot simultaneously compare two policies if one says to raise interest rates and the other says to lower them. In brief, our experiments are not repeatable. You can’t dip your toe in the same river twice. Thus, developing theories and policies to solve the many problems of the world has proven impossible—so far.
A new approach to understanding theory is needed because we have an overly large number of problems in this world, and a great shortage of useful theories that we might apply to solve them. Such an approach should be rigorously scientific and should also pursue a different path from those that have already been followed. The study of conceptual systems may also be considered as an important direction because it supports the development of other sciences—especially the social/behavioral sciences—which so far have not proved highly useful.
Recently, there is increasing interest in the topic of theory creation. Indeed, there have been a number of useful and insightful texts discussing important aspects to building theory (M. Edwards, 2010; M. Edwards, G., 2009; Shoemaker, Tankard Jr., & Lasorsa, 2004; Van de Ven, 2007; Wallis, 2010b).
While we do have plenty of data, part of the problem (and a new direction), may be what we plan to do with that data. Shaw and Allen (2012) note that theory construction becomes problematic when the conceptual components are chosen in a way that is not systematic. In general, those who develop theory, “rarely build their theories in textbook fashion from the bottom-up. Real theory building is less organized and a good deal more fun. You start at the top, with an intuition for how some systems, some structure of things or concepts works. From there you feel your way, by intuition trial and error luck and logic, to what looks like the right answer” (Friedman, 1997: 40). This approach is far from scientific or systematic. It should come as no surprise, then, that our theories are of rather limited usefulness in understanding our systemic world. In short, a non-systemic approach to building theory has not led to the creation of useful theories. So, we may benefit from a more systemic perspective.
While concepts are already considered important within the science of cognitive systems, conceptual systems are too often thought of in terms of language and its analogical relationship to the world (e.g., Gentner, 1983; Johnson-Laird, 1980). Again, an approach that is not highly rigorous.
At the same time, within the systems sciences, interdisciplinary thinkers have worked to understand social systems, biological systems, chemical systems, and so on. A gap in understanding exists because comparatively little has been written about our systems of concepts (which may be understood as a set of interrelated ideas, concepts, or propositions). Such conceptual systems are in contrast to (for example) biological systems (consisting of interconnected cells) or a chemical system (consisting of interconnected atoms). Examples of conceptual systems include theories, models, mental models, policies, and others.
Although the social/behavioral sciences have worked to systemically understand our world, we have not made the same rigorous effort to understand our conceptual systems. This is strange because it is those conceptual systems that we use to understand and engage the world.
Therefore, in taking a more purposefully scientific and systemic approach, we will begin by defining a conceptual system more specifically as a set of interrelated propositions. Examples of those propositions may be found in theories, models, schema, conceptual maps, strategic plans, and policies. Any place where a concept is described in relation to another concept. Mental models may also be useful data for analysis, particularly when surfaced through writing and/or speeches. On the edges of the science, valid data may also include partial systems such as individual propositions and hypotheses. Such partial systems are useful in striving to understand the “building blocks” of larger systems.
The process of this science is to quantify aspects of conceptual systems; and, importantly, to find links between the measures of those systems and their usefulness in the world.
In the classical approach to science, we live in a “world of data.” So, theories are evaluated on the strength of their data. That is to say, the correspondence between concepts and reality. In contrast, this emerging science works under the assumption that we live in a “world of systems.” And, such a world will be better understood by theories which are more systemic. So, theories are evaluated for their internal coherence.
Coherentist perspectives have been applied to evaluate scientific revolutions. And, have arguably provided better explanations (Šešelja & Straßer, 2014). Naturally, we need both coherence (interconnecting logics within a theory) and correspondence (empirical evidence or facts relating to a theory) in order to have a useful theory.
In this paper I will provide a brief history of this emerging science of conceptual systems, explain some methodologies, their sources of data, and the understandings that they have generated. I will also provide suggestions for extending the science-based research in a variety of directions. Essentially, I am seeking to open a conversation that asks how this line of research might be extended to gain new insights—and eventually develop more useful and generally accepted methods for creating and evaluating theory to generate theory that is more effective in practical application to understand and resolve the problems of the world.
The science of conceptual systems is set within the broader science of cognition begun in the 1950s. Cognitive science is focused around the investigation of mental representations, mental processing of reasoning and inference, and the meaning of words and phrases (Johnson-Laird, 1980). However, where cognitive science investigates the more complex issues of thinking, the present science is focused on the structures that result from, and/or guide, the thinking.
From Aristotle’s “pictures in the mind,” we have understood that our minds contain some kind of image or understanding of the world. Because there are some things that cannot be pictured (especially abstract notions such as freedom) Kenneth Craik (1943) began talking about a “mental model” instead of a picture. This was a subtle, through significant shift because having a model implies having some structure. Which, in turn, suggests that it may be amenable to analysis.
This was not to be an easy path of investigation. Lane (1992, drawing on Lakoff and others) notes that, “there is overwhelming evidence that we humans do in fact maintain overlapping and inconsistent conceptual systems” (p. 27). It also appears that we don’t do much better when transferring those conceptual systems to paper. Even our ideas of philosophy, “are often themselves a contradictory and confusing patchwork of fragments of various philosophers’ representations of the world” (Ledoux, 2012: 14). In short, in our minds and in our writing, we have a great deal of potential data—but it is rather jumbled. We need rigorous analysis to make sense of it all.
The path continued when Kelly (1955) suggested that mental models must have a kind of structure to them. Thus bringing us a step closer towards scientifically rigorous analysis.
Over the years, a number of tools have been used to focus on the “internal” aspects of texts. These include scientometrics and bibliometric approaches to quantify the content of theory (c.f. Calas & Smircich, 1999; Fiske & Shweder, 1986; Flower & Mellon, 1989; Jean-Pierre & Edward, 2000; Knorr-Cetina, 1981; Kostoff, del Rio, Humenik, Ramírez, & García, 2001; Ritzer & Smart, 2001; Zimmer, 2006). Some related work may also be found in formalizing structures of sentences (Gentner, 1983) such as formal language or “natural language.” More attention should be paid to using and developing the tools from these areas. However, these approaches have not purposefully addressed the systemic structure of theories.
An important step forward occurred when, building on Kelly’s insights, scholars developed Integrative Complexity (IC) as a method to analyze the structure of mental models (Suedfeld, Tetlock, & Streufert, 1992) using the paragraph completion test.
Briefly, IC was used to evaluate paragraphs of text (from writing samples, speeches, government announcements, etc.) and rate them on a scale of one to seven. A score of one indicates a low level of interrelatedness between the ideas of the text and a score of seven indicates a high level of interrelatedness. For example, a simple statement of universal truth would receive a very low score. In contrast, a complex statement (as may be found in many academic and philosophical writings) would receive a high score.
Studies using IC found that higher scores were significantly correlated with managerial effectiveness (Wong, Ormiston, & Tetlock, 2011), political success (Raphael, 1982), and higher student scores (Curseu, Schalk, & Schruijer, 2010). This ongoing path of research has shown that we can indeed evaluate the structure of our conceptual systems in a way that is rigorous and scientific. And, importantly, those conceptual systems with higher levels of structure are more useful and support success. Therefore, we may be guided into developing conceptual structures with higher levels of structure so that we may learn to more effectively address the problems of the world.
Figure 1 Relationships between some sciences.
Because that approach includes rigorous methodology and measurement (IC), data (evidence in the form of text), I suggest that IC represents the start of a science of conceptual systems. Further, that this new science exists at the intersection of cognitive science and systems science (Figure 1).
Before long, it was generally agreed that theories must also have some kind of structure (Dubin, 1978; Kaplan, 1964; Stinchcombe, 1987). Studies into the structure of theories seem to have great promise because we are studying the internal coherence—comparing concepts to concepts instead of comparing concepts to non-conceptual objects (Wallis, 2014a). That is to say, to learn more about apples, we should compare them with other apples. We can’t learn much about apples by comparing them to concepts (e.g., freedom).
However, little was done to quantify the structure of those interrelationships. In response to this gap, Integrative Propositional Analysis (IPA) was developed specifically to provide rigorous and quantitative analyses of conceptual systems such as academic theories and policy models.
By way of background, IPA is derived from Reflexive Dimensional Analysis (RDA) which was founded in Grounded Theory and dimensional analysis (Wallis, 2008). IPA also includes aspects of scientometrics and bibliometrics (Wallis, 2009a), as well as structuralist perspective (Wallis, 2009b), integral thinking (Wallis, 2010d), systems thinking (Wallis, 2010c), complexity theory (Wallis, 2011), and narrative analysis (Wallis, 2013b).
IPA begins by identifying specific conceptual systems for analysis. Within each (theory, policy, model, mental model, etc.), causal propositions are then identified. Each proposition includes one or more “things” and the relationships between them. Things are expressed as concepts which may be as concrete as (dare I say) “concrete” or abstract as “abstraction.” Between things, causality is the best way to describe specific relationships.
Johnson-Laird (1980) neatly explains the importance of causality for improved understanding (p. 107). And, it has been shown that maps based on causal relationships have been used effectively in business settings (Axelrod, 1976). In short, causality is important because it is the best path to scientific understanding (Pearl, 2000).
While it is generally held that theories, and their propositions, are best when derived from observational data, creative insights may also be useful—perhaps even necessary (c.f. Ambrose, 1996; Stacey, 1996; Thagard & Stewart, 2011; Uzzi & Spiro, 2005).
The structure of those propositions may take many forms (e.g., atomistic, linear, circular, branching). While there are varying degrees of usefulness to those structures, the one that seems most useful to date is the “concatenated” logic structure (Wallis, 2014c).
Figure 2 Simple concatenated logic structure.
In Figure 2, A, B, and C represent concepts and the arrows represent causal connections between them. Specific changes in A and/or C (e.g., increase or decrease), will cause changes in B.
IPA is a six step process:
Using Figure 2 as a very simple example of an IPA analysis, one may identify three concepts (A, B, and C). Therefore, the Complexity of the theory is C=3. Among these, there is one concatenated concept (B). Dividing the concatenated concept by the total concepts gives us a Systemicity of 0.33.
One might say that the Complexity of the conceptual system is a measure of its “breadth” while Systemicity is a measure of its “depth.”
We can use IPA to analyze text where causal relationships are described. Useful sources of this kind of data include academic journals, books, policy statements, and similar material. The results of such studies may be presented in a variety of ways to generate new insights into the subject matter. In the following section, we will look at the results of some recent research. In each of the following figures, each dot represents the analysis of one theoretical model drawn from the literature.
One type of study is to use IPA to evaluate changes in Systemicity of a theory over time. Here, the goal is to identify macro trends, rather than attempt to make fine-grained inferences about the efforts of individual theorists. Although, of course, the fine-grained approach represents an interesting direction for additional research.
Figure 3 shows the evolution of a theory from ancient times (low Systemicity and low level of usefulness) through the scientific revolution (rising Systemicity and rising level of usefulness), through its development into a “law” (perfect Systemicity and high level of usefulness).
In using Complexity and Systemicity to choose one theory from among many, the first choice would be to use the theory with the highest level of Systemicity (greater usefulness and greater depth of understanding). The second choice would be to use a theory with higher Complexity (more concepts and greater breadth). By using a theory with higher level of Complexity, one may expect some increase in usefulness (including explanatory and predictive power). There is a cost, however. Engaging a theory with a large number of concepts means that one should track data from all those many concepts—often a difficult task. Realistically, when choosing from a range of theories, a scholar might be forced to compromise between a theory with higher Complexity and one with higher Systemicity. The important thing is that such a choice should be made consciously. It (almost) goes without saying that those theories are best when supported with empirical studies and include concepts related to the topic of interest.
Figure 3 Systemicity of electrostatic attraction theory (adapted from Wallis, 2010a).
Understanding the evolving level of Systemicity is useful because it provides an alternative view of the scientific revolution. Where traditional views suggest that the revolution was due to changing metaphors of electricity and increased reliance on empirical data, this new perspective suggests that the success of the revolution was also due to a rapidly increasing understanding of how the empirical data fit together with specific structures of logic.
This is an important shift of perspective because it provides a more objective evaluation of the structure of a theory—along with a more reasonable goal of creating better theory by developing a measurably higher structure. In contrast, the idea that a “change of metaphor” might be causal to a scientific revolution may lead some to believe that any change of metaphor is tantamount to a revolution in science. Indeed, many publications claim that their work represents a paradigmatic shift—when the actual changes to the science are not so impressive (Clegg, Cunha, & Cunha, 2002).
Figure 4 Complexity of electrostatic attraction theory (adapted from, Wallis, 2010a.
Another result of IPA highlights the Complexity of theory. Figure 4 shows the same theory over the course of its evolution. This time, however, we are looking at the Complexity (the number of concepts) within the theory. Interestingly, the theory became more complex during the revolution. This seems to have occurred as scientists created larger theories to explain new and different ideas. While it may prove a difficult mountain to climb, ascending a Complexity peak of this sort may provide a positive course to advancing theory in any science.
Figure 3 and Figure 4 both show very little change over the centuries before the scientific revolution. This suggests that the longest-lived theories are not necessarily the best.
More comparative studies may be conducted with other theories from the natural sciences. The results are expected to be similar. This is because those theories all began with a low level of usefulness and ended with a high level of usefulness. And, similarly, they ended up being amenable to algebraic manipulation—which is much the same thing as having a Systemicity of 1.0.
Within the social/behavioral sciences, it is difficult to compare such theories as those sciences generally present us with large numbers of theories—with little confirmable usefulness. For example, Figure 5 presents a set of well-known theories of conflict from sociology (drawn from Turner, 1986). Here, we see that the level of Complexity of those theories appears to be in slow decline. Similarly, in Figure 6, the Systemicity of those theories is also in slow decline. There is no evidence of a revolutionary increase in the Systemicity of these theories. And, outside the ivory tower, we continue to face the tragedy of war. So, it seems that our ability to understand and prevent war is also not improving.
Figure 5 Complexity of theories of conflict from sociology (adapted from Wallis, 2015, Under submission).
Figure 6 Systemicity of theories of conflict from sociology Wallis, 2015.
I will resist speculating on the events of history, or the personal lives of the theorists, that may have impelled changes in these paths of evolution. However, such directions of thinking do indicate new directions for research that is more fine-grained than is presented here. It is beyond the scope of the present coarse-grained analyses to suggest (for example) that the outbreak of war impelled a theorist to new insights. Or to revise a theory to higher levels of Complexity or Systemicity. Similarly, additional analyses might investigate changes in theory as developed by a single scholar—regardless of world events. For example, does the theory become less Complex and more Systemic as the scholar develops a deeper understanding (which may be on a “personal” level and might not be completely represented within the pages of the literature)?
In Figure 7, a similar study (this time with our data consisting of theories drawn from the highest cited paper of each decade) shows that theories of motivation are increasing in Systemicity. However, that increase is unsteady and does not show promise of reaching high levels of Systemicity any time soon.
Figure 7 Systemicity of theories of motivation from psychology (adapted from Wallis, 2012).
As may be seen in Figure 8, the Complexity of those theories is likewise unstable. However, here the Complexity is declining.
Here, again, it is beyond the scope of the present paper to explain or speculate usefully on the events (historical and/or personal) which may have influenced these changes. There are simply too many variables and too little understanding. Such explorations should be encouraged among interested scholars.
These studies provide a new view into the evolution of theories. By charting the history of a body of theory on this large-grained scale, we may gain a sense of where it has been, where it is going, and how long it may take to get there. However, as may be inferred from the relatively small sample size, they are only early explorations. Future studies should be conducted using a statistically relevant number of theories as data.
The main point to be conveyed here is that “something is happening” that is deserving of more study. And, importantly, this new perspective suggests a direction for gaining new insights into how those world events support the improvement of theory.
Figure 8 Complexity of theories of motivation from psychology (adapted from Wallis, 2012).
One area where we have been able to compare differing conceptual systems of the social sciences is in the study of policy models. By conducting a case comparison of policy models we have found that those with higher Complexity and Systemicity tend to be more effective in practical application. That is to say, they are more likely to be useful in helping nations to reach their goals (Wallis, 2011). Or, to achieve higher goals at lower cost.
Figure 9 shows the Systemicity and Complexity of two policies. The policies were drawn from the websites of two political parties before a recent election. IPA provides a new perspective for comparing these policies. Where individuals may be swayed by partisan rhetoric, this analysis suggests a strong argument in favor of adopting the policy with the higher Complexity and Systemicity because that policy is more likely to be useful and successful.
Figure 9 Comparing two economic policies as predictor of success (adapted from Wallis, 2013a).
Another form of analysis is presented in Figure 10. This data set is of nine theories of entrepreneurship (Wright & Wallis, Under submission). This scatterplot diagram may be understood as taking a “snapshot” of a set of theories within a field or sub-field. To obtain this sample, we surveyed Google Scholar ™ to find a set of theories that were highly cited (more than 200). Those six theories were supplemented by three more (a sample of convenience). We were surprised to find that theories with fewer citations (under 100) were also those with the lowest scores of Complexity and Systemicity (found in the lowest left-hand corner of the diagram). This finding hints at a new path for improving one’s citation rate in the academic literature.
Figure 10 Theories of entrepreneurship (from data in Wright & Wallis, Under submission).
For the Complexity, note how most theories are at C=10 and below. Like Figure 5 and Figure 8, the Complexity of theories seems constrained. This may be due to a number of influences—all of which are deserving of additional investigation if we are to advance our theories and our sciences more rapidly. One possibility is the simple limitation of space in journals. Another is the difficulty faced by scholars in collecting and presenting such a large number of concepts. In some sense, those may be considered perfectly reasonable explanations. However, if our goal is to create more effective theories and advance the social/behavioral sciences into something that is more useful, then we have a problem with the way we apply our social/behavioral science analyses.
For the Systemicity, in Figure 10, please note how no theory exists above the 0.4 level—and most are much lower. Like the theories presented in Figure 6 and Figure 7, the social sciences do not seem to be producing theories with a high level of Systemicity. Even though that seems to be required to develop theories that are highly useful in practical application—based on past philosophical perspectives (Dubin, 1978; Kaplan, 1964) and more recent analyses (reflected in the present paper).
In short, in Figure 10, theories are clustered in the lower left-hand quadrant. This suggests a valuable opportunity to identify (and/or create) theories in the other three quadrants. Perhaps theories inhabiting other quadrants may prove more useful in solving the problems of the world.
As you may well infer from the relatively small sample sizes presented in the above figures, this science is still in an early stage of exploration. Naturally, we cannot readily extend these results to suggest that all theories of the social sciences are in such poor condition. That kind of conclusion must wait upon more research. There is a vast opportunity to map the “theory genome” or create a “periodic table of theory” and gain new views into our world of theory across the social/behavioral sciences—and beyond.
These results lead us to a few key ideas about the structure of theory within the social/behavioral sciences. First, is that “parsimony” (although popular) is not (by itself) a useful or viable goal in theory construction. Dubin (1978: 316), for example, notes that the test of parsimony is often applied and misapplied. Meehl (2002) argues brilliantly against the claim of parsimony. Yet, others claim that “the simplest theory is the best” (Shoemaker et al., 2004: 172) and that there is a tradeoff between theories that are “general,” “accurate,” and “simple” (Van de Ven, 2007).
The comparison in the present paper suggests that the relationship is not so simple. See especially Figure 3 and #4 which show that a theory with low Complexity and low Systemicity is not useful. While, in contrast, a theory with a high level of Systemicity coupled with a low level of Complexity is useful. There also exists a middle ground where a theory with a high level of Complexity and a medium level of Systemicity will also have some level of usefulness. In short, at a minimum, scholars should avoid claiming validity for their theories based only on parsimony because it appears that Systemicity is a more accurate indicator for usefulness.
Certainly, a move toward larger theories will cause some difficulties—such as the limits of human understanding. However, the benefit of creating more effective theories means that these issues should be addressed, not avoided. One way to address the difficulty of engaging more complex theories is to encourage more collaboration between scholars. If each scholar were to focus on a set of concepts within a theory, their efforts could be coordinated along the lines of the causal connections between the concepts. This kind of effort would also support linking scholars for interdisciplinary efforts.
Another difficulty is the space available for publishing in journals. This might be alleviated by providing links to online content. Naturally, all of these efforts will require more coordination—and that suggests the need for more funding.
An additional idea is that the social/behavioral sciences should be striving to identify more causal connections between concepts within the model. Metaphorically, this approach provides a useful “compass” to guide researchers towards developing better theory through research. It also provides a new and more useful way to focus the efforts of researchers. Where Lakatos (1970) suggested that theories have a core of unchanging ideas, another researcher has thrown that into question—and suggested that the core of a theory is the set of concepts with a higher density of causal connections (Wallis, 2014b). Thus, researchers may coordinate themselves by identifying whether their work involves adding new concepts to the belt of the theory or adding casual connections to strengthen the core.
Generally, and metaphorically, this paper suggests that we should choose to create and use roadmaps with many dots connected by many lines; not maps with many disconnected dots.
Using the tools and perspectives of a science of conceptual systems, there are many opportunities to accelerate the advancement of science. Future studies may be conducted in at least three general directions. The first is to deepen the existing studies. Where the above studies were conducted with only a few theories as data, they should be replicated with hundreds of theories. In such an endeavor, we have no lack of data because each publication essentially presents a new theory. Although many scholars use the same title to refer to their theories, I can’t recall finding two identical theories—although many scholars will use the same title for theories built of differing concepts.
Theories chosen for these kinds of study should not be limited to those which are most cited or most popular. A more useful approach is to choose them according to more objective criteria—such as the strength of their underlying empirical research. Another approach is to draw upon thousands of theories, chosen at random and compare those theories along multiple dimensions over time (Faust & Meehl, 2002; Meehl, 1992, 2002, 2004).
Between the present studies (drawing on a few theories) and future studies (involving thousands) researchers drawn to this emerging field might consider analyzing theories found in existing meta-studies, benchmarks, or reviews of theories—such as the one used above for theories of conflict (Turner, 1986).
Second, the research stream could be broadened to include new areas. For example, comparing theories from all of the sciences. Here, of course, all sciences may provide useful data in the form of theories and their constituent propositions. As a more concrete example within the field of business, an effective study could analyze a wide range of strategic plans and/or mission statements. Naturally, such a study would need to control for many confounding variables. Was the business environment as stable as expected? Where the plans implemented properly? How much research went in to the creation of the plans? How many changes occurred before a new formal plan was deemed necessary? Studies of this sort might begin using a case-comparative research method in tandem with IPA (e.g., Wallis, 2011).
A third area for advancement is to experiment with conceptual systems in practical application. We need to take our theories and put them to use. This is an excellent opportunity for the growing number of emeriti professors. They might, for example, provide advice to their local city councils and county boards based on theories with a high level of Systemicity.
The science of conceptual systems (in general) and IPA (specifically) have the potential to advance science and practice in a number of ways. Potential projects may include:
With more effective ways to create and evaluate theories, we may be able to create a social-political system that is as effective and useful as our sciences. At least one effort, currently under development, aims to evaluate bills before congress using IPA and other methods: http://scipolicy.org/ Another group is striving to integrate theories across multiple branches of science using IPA and other approaches (McNamara & Troncale, 2012). And, a dissertation used IPA to gain new insights into US policy (Shackelford, 2014) noting the importance of complexity theory and this new methodology, “are vitally important to the field of public policy” because we need to methods to adapt to an increasingly complex and nonlinear world. Indeed, “Change makers must be able to change the way they play the game” (p. 146).
The history of science clearly shows that theories come and theories go (Kuhn, 1970). Indeed, in the social sciences, it has been suggested that 90% of the theories rise very rapidly, only to disappear with equal speed (Oberschall, 2000). So, it is reasonable to expect that IC and IPA will also evolve, change, and (hopefully) be replaced by something better. In this section, I will briefly suggest a few directions where progress and evolution might be encouraged.
First, Paul Meehl suggested an ambitious program of evaluating many theories over time according to many variables (Faust, 2005; Faust & Meehl, 2002; Meehl, 1992, 2002, 2004). The results of such studies can be expected to suggest additional insights into the creation of more useful theories.
Second, would be to explore the idea that theories might be "nested" within one another. Such an approach might open new directions for thinking about theories from physics to sociology. For an explanation, by way of comparison, let me note that IPA, so far, has looked to see if theories might be “connected” based on overlapping concepts. However, there may be no direct analogy between two theories and so no direct connection. There may, however, be “emergent properties” (Robertson, 2014). Emergent properties have been studied between physical, biological, and social systems. However, they have not been studied in conceptual systems. Here, it may be speculated that emergent properties on one level may provide links to theories on another level. For example, theories of emotion do not seem to directly relate to theories of evolution. However, emotions may guide actions which in some way influence evolution. Alternatively, theories built of more concrete concepts might be nested within theories built of more abstract concepts. These are rich areas for study starting on the philosophical level.
Third, investigations might study and quantify “causal loops” as structures within theories. Loops have already be used to business organizations (Senge et al., 1994). Indeed, it is possible that loops might even be better than concatenated structures as markers to quantify the usefulness of theories. An idea that should make cyberneticists happy (c.f. Dent & Umpleby, 1998; Yolles, 2006).
Finally, as discussed above, IPA requires causal relationships between concepts. However, IPA is neutral as to what form those causal connections might take. Alternative understandings of causal connections should be studied. For example, the deep and thoughtful work of Kent Palmer (2014) investigates the importance of causality from multiple perspectives. He notes how Aristotle describes four kinds of causality (formal, material, efficient, and final) which seem to relate closely to Kant’s categorical and hypothetical imperatives. It seems rather important to identify causes that are “orthogonal” to one another as a way to better understand the structure of our theories.
For a concrete example, it is easy to say that planting more seeds will lead to more trees. It is more interesting to identify that the “seeds” and the “planting” are very different things—perhaps orthogonal to one another so that a better theory of tree growth would be developed by understanding “planting” and understanding “seeds” both separately and in conjunction. If we can create new rules for evaluating theories based on the orthogonality of causal inputs (and outputs), we will have made a very large step forward in understanding how to create more useful theories.
On a slightly speculative note, it is interesting that the above studies find so many theories with a Systemicity of about 0.25. This is curious because there are a number of studies suggesting that our organizational change methodologies tend to be effective around 20-25% of the time (e.g., Dekkers, 2008; MacIntosh & MacLean, 1999; Smith, 2003). Is this a coincidence? Or, is it possible that we might be able to double the effectiveness of our theories, practices, and policies by doubling their Systemicity?
The emerging science of conceptual systems is located at the overlap between cognitive science and the systems sciences. This paper has shown that the science of conceptual systems exists, not only as a philosophical curiosity, but rather as a tool to support other sciences so that they may advance more rapidly. Thus, the science of conceptual systems is not merely an “education multiplier” (e.g., teaching better ways to teach) it is a “science multiplier” (where advances in one science will impel advances in other sciences).
While it has been recently suggested that the advancement of the social/behavioral sciences is curtailed because we do not have an effective science of metatheory (Wallis, 2010b), the more recent work within the science of conceptual systems suggests that we now have the potential to make more rapid advances.
The approaches presented here do not provide a “map” for advancing the sciences. Indeed, none is available because we are advancing across that new terrain. What it does provide is a compass suggesting a new and potentially exciting direction to travel. Historically/traditionally, scholars have travelled about the countryside of the social/behavioral land making useful observations and gathering interesting data. They have often staked claims regarding the importance of their terrain. They could not advance more purposefully because they were guided only by intuition. So, they did not know what direction was “forward” although it seemed to have something to do with empirical analysis. Today, with the creation of a new science, we have a new direction
Another benefit of this approach is in counteracting the forces of fragmentation. Where increasing specialization has led to an explosion of new fields and sub-fields and all their associated theories, we are overwhelmed by their sheer number and variety. The science of conceptual systems offers paths for rigorously integrating those theories and connecting our fragmented fields (Wallis, 2014b). In short, where we see theories as data, our weakness becomes our strength.
However, IPA does not have an extensive track record. More studies are needed before we can have great certainty about its efficacy. Additionally, there are clear and interesting paths for developing alternative measures of conceptual systems. These should be investigated philosophically, qualitatively, and quantitatively.
History shows that theories emerge, evolve, and are replaced. It is very likely, therefore, that the same path of development will occur with IPA. In order to develop an improved version of IPA, it may be useful to use IPA in a wide variety of situations for research and evaluating the effectiveness of conceptual systems. That way, we can “push it” to find what its natural limits of usefulness might be. A world of scientific exploration awaits.
Council, S. (2010). Defining science, link.
Dent, E. B., & Umpleby, S. A. (1998). Underlying assumptions of several traditions in systems theory and cybernetics. In R. Trappl (Ed.), Cybernetic and Systems '98 (pp. 513-518). Vienna, Austria: Austrian Society for Cybernetic Studies.
Kostoff, R. N., del Rio, J. A., Humenik, J. A., Ramírez, A. M., & García, E. O. (2001). Citation mining: Integrating text mining and bibliometrics for research user profiling. Journal of the American Society for Information Science and Technology, 52(13), 1148-1156.
Lakatos, I. (1970). Falsification and the methodology of scientific research. In I. Lakatos & A. Musgrave (eds.), Criticism and the Growth of Knowledge (pp. 91-195). New York: Cambridge University Press.
McNamara, C., & Troncale, L. (2012). SPT II.: How to find and map linkage propositions for a GTS from the natural sciences literature. Paper presented at the 56th Annual Conference of the International Society for the Systems Sciences (ISSS), San Jose, CA.
Shotter, J. (1994). Conversational Realities: From Within Persons to Within Relationships, link.
Shotter, J., & Tsoukas, H. (2007, 7-9 June 2007). Theory as therapy: Towards reflective theorizing in organizational studies. Paper presented at the Third Organizational Studies Summer Workshop: 'Organization Studies as Applied Science: The Generation and Use of Academic Knowledge about Organizations', Crete, Greece.
Suedfeld, P., Tetlock, P. E., & Streufert, S. (1992). Conceptual/integrative complexity. In C. P. Smith (Ed.), Handbook of Thematic Content Analysis (pp. 393-400). New York: Cambridge University Press.
Sussman, S., & Sussman, A. (2001). Praxis in health behavior program development. In S. Sussman (Ed.), Handbook of program development for health behavior research and practice (pp. 79-97). Thousand Oaks, CA: Sage.
Umpleby, S. (2010). From complexity to reflexivity: The next step in the systems sciences. Paper presented at the Cybernetics and Systems 2010, Vienna, Austria, link.
Wallis, S. E. (2008). From reductive to robust: Seeking the core of complex adaptive systems theory. In A. Yang & Y. Shan (Eds.), Intelligent Complex Adaptive Systems (pp. 1-25). Hershey, PA: IGI Publishing.
Wallis, S. E. (2009b). Seeking the robust core of social entrepreneurship theory. In J. A. Goldstein, J. K. Hazy, & J. Silberstang (Eds.), Social Entrepreneurship & Complexity. Litchfield Park, AZ: ISCE Publishing.
Wallis, S. E. (2010a). The structure of theory and the structure of scientific revolutions: What constitutes an advance in theory? In S. E. Wallis (Ed.), Cybernetics and systems theory in management: Views, tools, and advancements (pp. 151-174). Hershey, PA: IGI Global.
Wallis, S. E. (2012, July 22-27, 2012). Theories of psychology: Evolving towards greater effectiveness or wandering, lost in the jungle, without a guide? Paper presented at the 30th International Congress of Psychology: Psychology Serving Humanity, Cape Town, South Africa.
Wallis, S. E. (2013b). A systems approach to understanding theory: Finding the core, identifying opportunities for improvement, and integrating fragmented fields. Systems Research and Behavioral Science Journal (in press).
Wallis, S. E. (2015). Are theories of conflict improving? Using propositional analysis to determine the structure of conflict theories over the course of a century, link.
Wong, E. M., Ormiston, M. E., & Tetlock, P. E. (2011). The Effects of Top Management Team Integrative Complexity and Decentralized Decision Making on Corporate Social Performance. Academy of Management Journal, 54(6), 1207-1228.
Adapted (and extended) from (Wallis, 2011: 99-124)
❏ Aspect See “Concept.”
❏ Atomistic Logic A kind of logical structure found within a proposition that is reductionist such as “A is valid” or “A is true.” Or, more concretely, “Apples are important.” These would score very low in a study of Integrative Complexity.
❏ Bibliometrics Methods such as content analysis and citation analysis, typically used in library sciences, for the quantification of text from academic literature.
Branching Logic A logical structure found within causal propositions including three or more concepts where a change in one concept causes change in two or more other concepts. For example, a Branching proposition might say that changes in A will cause changes in B and C. For a more concrete example, “More teamwork will lead to more cohesion, and more results, and more frustration.
❏ Case Comparative Study Investigating and comparing two or more cases along multiple dimensions to draw insights and inferences. A useful start to building new theory (Eisenhardt & Graebner, 2007).
❏ Causal Relationship (causality) Where two or more concepts are related so that a change in one causes a change in one or more others. A causal relationship is often expressed as a proposition, hypothesis, or a diagram. A causal relationship such as, “More A causes more B.” may also be used as a general term in place of other more specific terms. Instead of saying “more” other indicators might be “better” or “less” (for example). Similarly, instead of “causes” more specific indicators might include such terms as “creates,” or “engenders.” In any case, the description must be specific to be valid. It is not useful to state (for example) that “A and B are interrelated” or “More A may cause more B” because the nature of the relationship is not causally defined. Logical structures often describe causal relationships (e.g., Linear, Branching, Concatenated).
❏ Circular Logic A logic structure where one cause leads back to itself such as Changes in A cause changes in B cause changes in C cause changes in A. Circular logics may be seen as feedback loops which are held to be very useful in understanding systems. However, they can also be misleading if researchers rely on too few loops to adequately represent the system. Then they may appear as tautologies (e.g., more A causes more A).
❏ Cognitive Science Interdisciplinary investigation of the human, social, and artificial processes around perception, information, reasoning, and decision making.
❏ Coherentist Perspective Coherence is focused on the relationships between beliefs. A statement is true/meaningful/useful based on its relationship with other statements (Šešelja & Straßer, 2014; Sosa, 2003). This is in contrast to a perspective of correspondence where a statement is held to be true based on its relationship to empirical data. The two perspectives are orthogonal and so most useful when used in combination.
❏ Complexity A measure representing the number of concepts within a conceptual system. This may also be understood as the diversity of ideas within a document. For an abstract example, consider a conceptual system containing the propositions: A is true; More B causes more C; More B causes more D. In such a model, there are four concepts (A, B, C, D). Therefore, the Complexity of the conceptual system is C = 4.
❏ Concatenated Logic A logical structure found within a causal proposition including three or more concepts where changes in two or more concepts cause change in another concept. For an abstract example, a Concatenated proposition might state that changes in concept A and concept B will cause changes in concept C. In that example, C is the Concatenated concept, while A and B exist within a Concatenated relationship but are themselves not Concatenated. For a more concrete example, “More collaboration and more shared goals will result in more teamwork.” Here, “teamwork” is the Concatenated (and better understood) concept.
❏ Concept The part of conceptual system that represents a concept, idea, or notion. The concept may be as concrete as in “apple” or as abstract as in “truth.” Concepts may be simple as in “numbers” or complex as in “left handed monkeys with undiagnosed trauma.” A concept is typically detectable, that is to say empirically measurable, but that is not an absolute standard. Concepts are part of propositions.
❏ Conceptual System A set of interrelated concepts. Examples include theories, policy models, mental models, schemata, etc. Metaphorically, they serve as lenses to aid in understanding and effective engagement of the world (Wallis, 2014a, 2014c).
❏ Correspondence Perspective Commonly associated with “normal” science or “Science One” (Müller, 2012; Umpleby, 2010) and the use of Toulminian logics, correspondence is focused on the relationship between a statement and some empirically verifiable existence.
❏ Grounded Theory An approach to theory development that is non-positivist, yet is solidly based on the use of data—although the data are generally qualitative. Data, often from interviews, are coded and categorized. Then, relationships are identified between the categories to identify a useful theory (Charmaz, 2006; Glaser & Strauss, 1967).
❏ Integral Thinking Understanding (or attempting to understand) the world from a transdisciplinary perspective where those many perspectives are interrelated.
❏ Integrative Complexity (IC) “Integrative complexity is a measure of the intellectual style used by individuals or groups in processing information, problem solving, and decision making. Complexity looks at the structure of one's thoughts, while ignoring the contents. It is scorable from almost any verbal materials: books, articles, fiction, letters, speeches and speech transcripts, video and audio tapes, and interviews.” link
❏ Integrative Propositional Analysis (IPA) Combined processes of qualitative and quantitative analysis involving rigorous hermeneutic deconstruction of propositions found in formal texts including the rigorous re-integration of propositions from those texts following a structured methodology. Also a process of meta-analysis for investigating conceptual systems to determine the Complexity of conceptual systems (diversity of concepts) and the Systemicity of the conceptual system (connectedness between concepts).
❏ Linear Logic A logical structure found within a proposition describing simple causal relationship between two concepts. Such as, “More A causes more B.” Both A and B exist in Linear relationship to one another. Here, A is the causal concept and B is the resultant concept). Linear structures can be of any length (e.g., More A causes more B which causes more C which causes more D… which causes more Z). For a more concrete example, “Having more shared goals leads to more teamwork which, in turn, leads to more productivity.” Within an explanation, this may also be phrased as, “A is true because of B and B is true because of C… because of Z.”
❏ Logic Model A set of interrelated logic statements such as a theory or a Policy Model describing causal relationships between the elements of the model.
❏ Mental Model A representation within one’s mind about how the world works. Useful for understanding and engaging the world and for making predictions.
❏ Metapolicy Generally, the study of policy—how the policy is created, applied, and evaluated. This may be rigorous as in the use of Integrative Propositional Analysis or fuzzier as in the use of historical narrative. May also be used to describe a policy on how to make policy.
❏ Metatheory Primarily the study of theory, including the development of overarching combinations of theory, as well as the development and application of theorems for analyses that reveal underlying assumptions about theory and theorizing.
❏ Model “A model is a simplified representation of a system at some particular point in time or space intended to promote understanding of the real system. As an abstraction of a system, it offers insight about one or more of the system's aspects, such as its function, structure, properties, performance, behavior, or cost.” link
❏ Narrative Analysis Qualitative approach for understanding how people make sense of their world based on spoken or written accounts of their experiences and how those experiences are interpreted.
❏ Parsimony Generally, the understanding that a theory is better when it is smaller. Or, as small as possible while including ideas that are necessary or useful. Ockham’s Razor is a common example.
❏ Policy Model A cognitive structure (like a theory) representing how a community or organization understands the world, thus enabling them to take specific actions to achieve their goals. As a sense-making structure, the policy model does not (strictly speaking) include goals or actions. Those are part of the broader policy.
❏ Proposition “A proposition is a declarative sentence expressing a relationship among some terms.” (Van de Ven, 2007: 117). For example, “More travel leads to more discovery.” (See Causal Relationship). Those terms are often understood to be “concepts.”
❏ Reflexive Dimensional Analysis (RDA) A process for creating a unified conceptual system from multiple conceptual systems through a process of categorization, abstraction, dimensionalization, and the identification of causal connections.
❏ Robustness An older term for “Systemicity”
❏ Schema Most commonly used in computer science and logics in referring to a formula and/or set of data used to represent a system.
❏ Scientometrics Approaches of Bibliometrics applied to study the development of science within and between fields.
❏ Social/Behavioral Sciences The purposeful study and advancement of understanding in all fields of human interaction including (but not limited to) psychology, sociology, policy, ethics, business, management, human development, organizational development, economics, and social anthropology.
❏ Systemicity A ratio describing the interrelatedness between concepts of a conceptual system on a scale of zero to one. Systemicity is calculated by dividing the number of Concatenated concepts by the total number of concepts in a policy (see Integrative Propositional Analysis). Systemicity is a measure of how well integrated the propositions of a conceptual system are, the degree to which they are understood as existing in a systemic relationship, and the level of causality between the concepts. Systemicity is also related to the effectiveness of the conceptual system in practical application.
❏ Systems Thinking Understanding that the world is made of systems, this interdisciplinary way of thinking explores how those systems effect one another. Systems thinkers strive to obtain a deeper understanding by looking at the connections of many things, rather than seeking to understand things in isolation.
❏ Theory An ordered set of assertions. Weick (1989: 517; drawing on Southerland).