My most recent post described the history of the periodic table in chemistry. Behavioral research today has problems similar to those faced by chemistry in the 1800s, with a need for (a) consistent classification of elements, (b) periodicity or the grouping together of related items, (c) underlying mechanisms that can explain the similarities among elements as well as their differences from one another, and (d) inter-relationships of the elements to one another. The periodic table accomplishes all of these things, and it doesn't have to be arranged into rows and columns to do it. Above is another alternate version of the periodic table that shows elements arranged in a circle, progressing clockwise outward from the center and showing groupings of related elements as colored rays. Behavioral science, though, is struggling with even the first of the periodic table's beneficial properties.
Classification. It can be hard to get behavioral scientists even to agree on what we study. A range of well-established theories exist to explain behavior, such as the Health Belief Model, the Theory of Planned Behavior, the Information-Motivation-Behavior (IMB) model, Social Cognitive Theory (featuring Bandura's concept of "self-efficacy"), and Kahneman's Nobel-prize-winning Prospect Theory. Many of these theories posit the same or related constructs, such as beliefs, knowledge, motivations, social perceptions, and emotional reactions. These elements then get mixed together in different ways to provide explanations for people’s behaviors. Despite the definitive-sounding names of the rival theories, our science has not advanced very far from the “naïve psychology” identified by Fritz Heider in the 1950s, the causal model used by the average person-on-the-street who can clearly tell you that anyone’s behavior is the product of things like their thoughts, their feelings, and their intentions. The “periodic table” analogy holds out hope that we can do better than this as a field of science.
My colleague Kai Larsen from the University of Colorado Boulder has suggested that behavioral science suffers from an unnecessary proliferation of concept names. He has further divided the issue into two separate problems: the "jangle fallacy" in which the same idea goes by various names in the literature (e.g., motivation, readiness, and stage of change), and the "jingle fallacy" in which very similar concept names mean very different things (e.g., usability and usefulness in technology acceptance research). As described above, the first stage in a valid taxonomy of behavioral science is to agree on a classification system for its fundamental elements. One proposed solution to these problems, promoted in particular by psychologist Susan Michie and colleagues, is to work with teams of experts to agree on standard terms. This is essentially a democratic approach, working with a group of knowledgeable people and a whiteboard to make lists and consolidate them. Sometimes a formal knowledge-development process like the Delphi poll method is used, while in other efforts the process of coming to agreement is more ad hoc. Michie and colleagues have specifically developed a list of behavior change techniques used in psychological research, with concepts like "consequences," "rewards," "repetition," "antecedents" (meaning environmental stimuli that can be changed to prompt different behavior), "social support," and "feedback." I tried my own process like this with CU Nursing faculty colleagues in 2011, coming up with a list of psychology and behavior constructs that was quite similar to Michie's. And an independent effort using the expert consensus method in an 81-study literature review of behavior change interventions had 86% agreement with Michie's taxonomy (interestingly, these authors included an "education" domain that Michie omits, perhaps based on the well-known finding that knowledge change does not lead to behavior modification). A third set of investigators in 2015 produced a set of tables summarizing 13 basic behavior change techniques with sub-techniques included under each one (full list available here), again with 77% agreement to Michie's list (exceptions were "tailoring," "individuation," and "motivational interviewing"; two of these are more like delivery strategies than actual techniques, while motivational interviewing is a therapy approach that incorporates several of the other techniques). The relationship between Michie's umbrella terms for behavior change techniques and others that have been used in the literature is shown in this cluster analysis diagram (also available as a PowerPoint slide here):
Kai Larsen himself proposed a very different approach to the problem, using natural language processing algorithms that allow computers to understand humans' use of words (think Alexa or Siri, but for writing). In a 2012 paper, some CU Nursing colleagues and I applied Kai's method to the concept "self-management" in the context of diabetes. Nursing uses an expert-consensus method called concept analysis to identify the meaning of scientific terms, including its antecedents, attributes, and consequences. We tried Kai's algorithm-based approach and generated results that were close to those of the traditional expert consensus approach -- of 17 terms related to self-management that were identified through concept analysis, the INN natural language processing approach found 15 of them.
Overall, the research reviewed in this section suggests that we still have a ways to go simply in order to get researchers to agree on their use of terminology, the basic "elements" of behavioral science. Still, there are promising developments. Two Minds Theory currently includes a list of 5 different behavior-change strategies, which overlap 100% with Michie's organizing framework, although some of the items on my list like behaviorism include subsets of items that are on Michie's list individually. Michie's framework is both the best-studied and the most comprehensive of the currently available classification systems, although Larsen's novel method also has much to recommend it because machine learning algorithms may find aspects of constructs that human experts cannot. The more that different approaches and different groups of investigators produce the same lists (even if the words used to describe the concepts are slightly different), the more we can be sure that we are "carving nature at its joints" and identifying the true elements that make up behavioral science.
Periodicity. Beyond simply listing the concepts used in psychological research, Michie's approach holds out the potential for finding related groups of items. These are shown at the bottom of the cluster analysis diagram above. For instance, Michie's Group 12 incorporates self-belief, comparison of outcome, and identity. What all of these approaches have in common is that they speak to the notion of who one is, a fundamental sense of self that is expressed in emotional, cognitive, and social or interpersonal terms. A sense of self implies the presence and perceptions of other people, and may relate to the TMT concept of social imagination, one of the key functions of the Narrative Mind. Indeed, personal narratives seem to underlie all three concepts: beliefs about oneself (including self-talk, the things that you say to yourself about yourself), persuasion by others ("comparison of outcome" in Michie's model), and taking on a role or identity that defines one in society. These three items in Michie's taxonomy do seem related to one another, and there may even be an underlying biological mechanism (the Narrative Mind, centered in the prefrontal cortex) that can account for all three of them. Here's the list in periodic-table form:
What this representation suggests is that items in the group have some core similarity, like the similar chemical behavior of element groupings that were identified by medieval alchemists. Items in the diagram above differ across rows based on some other characteristic. The labels that I gave to the rows (periods) in this "periodic table" are purely hypothetical at this point, but I based them on one of the oldest models of mind from Heider's naïve psychology. Heider documented people's naturally occurring attributions about the causes of events, and determined that people most often said that people did something because they thought it was a good idea, because they felt like or wanted to to it, and/or because they intended to do it. All of these ideas are basic to our understanding of other people, and therefore are good candidates for cross-cutting categories that map to various psychological concepts.
Underlying Mechanisms. Underlying mechanisms are the basis for the common idea that "biology reduces to chemistry; chemistry reduces to physics"; in each case, even more fundamental properties of nature can be used to explain observable phenomena. There's a common perception that psychology ought to reduce to biology, but our ability to actually accomplish that reduction has failed on many occasions! Still, if we are looking for underlying mechanisms to explain psychological concepts, biology is probably the place to start. In the example above, there is some basis for thinking that Heider's folk psychology concepts result from underlying mechanisms that are based in human biology. For instance, some concepts are more cognitive, based in the cortex's language systems; these systems also handle logic and math, and seem to be responsible for what we would naturally call "beliefs" or "attitudes." These higher-brain functions are part of the Narrative system in Two Minds Theory. Some concepts are more emotional, originating in the limbic system in the lower brain; Two Minds Theory considers these functions to be part of the Intuitive system. In the third row, interpersonal functioning probably involves a mix of cortical and subcortical systems that help us get along with others. Intuitive thinking is sometimes characterized as a "set of systems" rather than a single system, and the social versus emotional components of the Intuitive mind might turn out to be separate neurocognitive systems just as the Narrative and Intuitive minds are. If the three rows in this table are in fact found to originate in different underlying brain mechanisms, then we would be much further along on our way to a periodic table of behavior because we would have found a fundamental principle like valence in chemistry that explains the observable phenomena that we see in studying behavior.
In a different area of science from Michie's behavior-change studies, a major theory advance of the past 25 years is the Five-Factor Model of Personality, which has largely replaced older theories of personality like Freud's or Jung's. This model proposes 5 basic traits that shape people's overall worldview and interactions with others: 1) introversion-extroversion, 2) agreeableness, which is how well you get along with others, 3) conscientiousness, which is your level of attention to detail, 4) anxiety or neuroticism, which is how much you worry about things, and 5) openness to experience, which describes whether you tend to go out and try new things or stick to the familiar. The five factors were originally developed by comparing and combining terms from multiple competing models of personality, so the first component of a periodic model (classification) is well demonstrated. In this case the researchers used factor analysis to mathematically identify like terms, so the approach is more like Larsen's algorithmic strategy for classifying behavior-change interventions than like Michie's expert-consensus approach. Even more importantly, all five of these traits have shown genetic associations, with particularly strong heritability for neuroticism and openness to experience. Specific gene markers have even been identified for agreeableness. These findings suggest an underlying biological mechanism that can explain the five-factor model.
For the second criterion, can the grouping of items in the five-factor model be organized according to some of the same basic concepts as I proposed for Michie's classification of behavior change methods? Here's how I might organize the five personality dimensions:
Overall Status. Overall, the science of behavior seems to be at a point similar to chemistry at the time Mendeleev developed the periodic table. We are working towards an elemental list of psychological concepts, which can be validated independently by different teams using different methods, even if different researchers use different words to name these concepts. For some of these concepts, there seem to be common properties that link them together as a group with similar properties. And in a small but growing number of cases underlying biological mechanisms seem to exist that can describe both the similarities and the differences between items in the lists. I'd like to think that Two Minds Theory -- the Narrative versus the Intuitive system, with their respective origins in different parts of the brain -- could be one such mechanism to help us understand the similarities and differences between items in a group.
What we don't yet have is any clear concept of the inter-relationships between items. How would we know, for instance, if the "personality" grouping above belongs to the left or to the right of the "behavior change" grouping in our proposed periodic table of behavior? As in the case of chemical research on atomic numbers, electron shells, and radioactivity, there are likely some fundamental properties of the brain or human experience that have so far eluded us. Still, the elusive idea of a "periodic table of behavior" remains something to strive for, something that will help us explain and predict behavior just as one can predict the outcomes of chemical reactions simply by knowing the basic elements involved.
Comments
Post a Comment