Christmas Trees, Ornaments, and Action Phases

A Building Block for Custom Behavioral Analysis

Jeff Brodscholl, Ph.D
Greymatter Behavioral Sciences

In the past (including in at least one earlier post), I've offered some opinions on behavioral science frameworks, most of which boil down to the idea that, while they can facilitate a team’s consumption of behavioral science ideas and lend some support to an applied scientist’s work, they really can't act as a full-blown substitute for a behavioral scientist’s knowledge or skill – not if the goal is to have the ability to apply the science in a way that can meet a broad array of challenges with the firepower they sometimes require.

Yet, if we’re not going to rely on individual frameworks to guide our work, then how can we possibly apply behavioral science efficiently and at scale given the time and resource constraints within which the work must be performed? My own answer has been to take an approach that is, oddly enough, slightly similar to one found in the Behavior Change Wheel, which is to say that we can get to where we need if we bring to any challenge a toolkit consisting of the following four sets of items, constituted properly so that we’ll be empowered to build as deep and holistic a picture of behavior as needed for the problem we’re being asked to address:

  • Set 1: A collection of behavioral models that can help us outline the journey people take to arrive at, and execute, a target behavior or set of behaviors within a particular context of interest
  • Set 2: A collection of theoretical concepts, representing behavioral drivers or processes, that we can append to the journey to further elucidate the determinants of current and desired behavior
  • Set 3: A toolkit of analysis and research tools that we can use to put the pieces from Sets 1 and 2 together in a logical, evidence-based way, and to test hypotheses that our efforts generate
  • Set 4: One or more processes for executing the work, which we can commit to before the work begins

Key here, though, is the way in which these sets are, in fact, constituted. Unlike the Behavior Change Wheel, I assume that what's in these sets is going to be much richer and more varied than what you'd ever get from a framework – the promise being that, by directly embracing the published literature in all of its complexity and building the sets to reflect that depth of knowledge, the result can be custom applications that will fit the requirements of any challenge while staying close to what it is that the science can tell us.

Alas, saying that we should “embrace the complexity of the literature" doesn’t mean we can go into an engagement with a free-for-all grab bag of things and expect to come out alive, either. At some point, we do need to give careful forethought as to what to bother including in each of our sets so that, when we encounter a challenge, we'll not only be sure to have something available that makes sense to apply, we'll also be confident that we'll know how to use it to good effect and not just be left flailing about.

In this post, I’m going to talk a bit about an item I keep in my Set 1 which I think fulfills on some of these requirements. This item, which is sometimes referred to as the “Rubicon Model” or “Mindset Theory” of Action Phases [1-5], comes from an area of behavioral science interested in how people "self-regulate" – that is, plan, weigh options, learn, and adapt – as they pursue their goals. I focus on a Set 1 item here as the decision regarding what to pull from that set plays an important role in determining the contours of any conceptual model we develop for a behavior we’re interested in, which is a key step in understanding the nexus of drivers influencing that behavior and finding the opportunities to intervene. I sometimes refer to the process of building such a model as being a little like decorating a Christmas tree, in which we:

  1. Start by choosing an item from Set 1, which gives us a bare-bones, high-level model of the behavioral processes that unfold through the journey to the action that interests us (this being the “Christmas tree” that we’ll need to decorate); and then
  2. Add items, or “ornaments”, from Set 2 to fill out the picture to the point where we have a model with sufficient complexity to fully appreciate the barriers to change and identify something that we can plausibly target for intervention

(The process is a little more complicated than that, given that I might need to look for “drivers of drivers”, which is like adding ornaments to ornaments, but you get the general idea.)

In dedicating this post to the Action Phases model, my goal is not to suggest that I see it as a magic bullet that can be touted as the solution to all insights and intervention challenges. That’s one of the reasons why I assume Set 1 is, in fact, a set, and not just a single behavioral model, such as COM-B or B=MAT, that aspires to universality. Ultimately, the decision as to what gets pulled out of Set 1 needs to be determined, in part, by the behavior in question and the context in which it unfolds: What makes sense for understanding driver distraction, and for designing a console that will put an array of functions at the driver’s fingertips while keeping the driver’s focus on the road, isn’t necessarily what’s going to make sense for understanding, say, how a person onboards with, and then becomes adherent to, a self-infused medication with a complex dosing schedule.

That said, the Action Phases model does a pretty good job of addressing cases where a behavior requires some form of goal consideration and commitment, as well as a certain degree of forethought and monitoring, even when some of the pieces are automatic, the behavior recruits from well-worn skills and habits, and the timescale is not as long as, e.g., becoming adherent to a new, complex medication. In that sense, the opportunities for application can go beyond the heaviest cases to even simpler ones requiring some amount of volition, making it worthwhile to have in the consideration set no matter what area of behavior change you typically focus on. As I’ll show, the model also fulfills on a key function, which is to provide an initial conceptual structure that has just enough texture to be properly and efficiently embellished to suit the needs of the challenge in front of us.

The Action Phases Model in a Nutshell

If you haven’t heard of the Action Phases model until now, you may still have seen one of its most significant offshoots in the form of a particular type of intervention. The intervention, which involves a simple exercise in thinking, goes something like this (using voter behavior as an example):

  • Do you have a plan for voting in the November 7th election? If not, then take a moment to develop a plan for voting now. Write out where, when, and how you will cast your vote in this year’s November 7th election. Then, make this your plan for voting this November 7th.

This “implementation intentions” intervention has been the subject of over 20 years’ worth of research into its ability to support behavior change across a broad range of contexts. What this body of research has shown is that, while the intervention has its limitations and boundary conditions [6-9], it is, on average, quite effective at facilitating behavior change relative to the amount of effort and cost it takes to implement [6-7, 9-13].

So, here’s a question: What made anyone come up with this intervention, and why does it work?

The answer lies with the Action Phases model, which, at its core, looks to understand how people go from intending X (e.g., voting) to actually doing X when we know that the strength of people’s intentions often isn’t enough to guarantee that the intended action will happen [14]. The model addresses this issue by offering two large puzzle pieces, one of which helps to organize much of what we know about the ingredients in people’s pursuit of goals (e.g., “go and vote so that my favorite candidate has the best chance of winning”), the other of which adds new insights into the psychological consequences of these goal pursuit processes.

Puzzle Piece #1: The “Rubicon” Model

The first puzzle piece, which acts as a core substrate to the Action Phases model, is a little like what advertisers try to capture in a marketing funnel such as “AIDA”, except with a twist: Instead of spelling out individual steps involved in persuading someone to take an action and then putting action at the end, the Rubicon model puts intention formation at the front and then focuses on the steps that bridge the gap from intention to action. What results is a picture of the journey to an action that looks something like this:

Post3_Graphic1_RubiconModel

Here’s how to think about this journey:

  • In the first “predecisional” phase, a person considers whether to pursue a particular behavior (e.g., “go out and vote this November 7th”). This entails thinking about the immediate and future pros and cons of pursuing the behavior versus doing something else, including doing nothing (e.g., “stay at home again”). It also entails thinking about whether follow-through will be feasible, and what the chances will be that the considered pros and cons will occur if, in fact, the behavior is enacted. The deliberations provide the answer to why the behavior in question makes sense to attempt: The stronger the “why” behind it relative to the alternatives, the greater the likelihood of pursuing it.
  • The second “preactional” phase happens once the intention to pursue a particular behavior is formed (e.g., “I will vote this November 7th”). This is the first of three “post-decisional” phases through which a person travels once they’ve “crossed the Rubicon” from thinking about doing something to acting on it. In this phase, consideration moves away from lofty ambitions to more granular thinking about the specific actions one can take to enact the desired behavior (e.g., “I’ll vote by casting an early mail-in vote”, vs. “– by showing up at my local polling station”). This phase may be short or skipped for behaviors that are simple or habitual, but will be longer and more effortful for more complex actions. In the latter case, thinking focuses on developing a concrete understanding of the “where, when, and how” of each considered action. Once an action is selected, these "where, when, and how" thoughts form the basis for an “action plan” which not only organizes later steps, but creates a mental image of action in context so that, when the context is encountered later, it can cue the action without requiring much effort. (The answers someone might give to the intervention described earlier would be one example of such a plan.)
  • It's then in the third “actional” phase where actions in service of the desired behavior are performed. This phase may proceed smoothly, or may require learning and efforts to cope with barriers and setbacks. In either case, feedback provides a mechanism to understand how well actions are going as they are being executed. If setbacks are substantial, they may not only prompt additional efforts to learn, cope, and course-correct, but also trigger a round of questioning about the wisdom of the original behavioral commitment. This can result in a return to earlier phases where one now considers whether to stay the current course (e.g., “I’m still going to get to the polls this morning”), swap expected actions for new ones while maintaining the original commitment (e.g., “I’m still going to vote, but maybe now I’ll to do it in the evening”), or simply disengage and go back to the status-quo (e.g., “I’m just going to stay home”).
  • Finally, the “postactional” phase occurs when the desired behavior is fully enacted (“I voted!”). Here, feedback is used to evaluate whether the actions yielded the outcomes they were hoped to achieve (e.g., “did my candidate win?”). Similar to the actional phase, evidence that the behavior failed to produce desired outcomes is used to determine whether the behavior should be committed to in the future, or should be modified, supplemented, or dropped in favor of something else (e.g., “I just won’t vote anymore”).

Notice that this journey helps explain why the “implementation intentions” intervention would be a good way to support behavior change: It essentially covers bases in the “preactional” phase by prompting anyone who’s attempting a new, complex behavior to think through the steps involved in making the behavior happen. It also helps to explain why the intervention would be so effective: It works to forge a mental connection between intended actions and contextual cues, facilitating the transfer of the behavior, or the remembrance of the intent to perform it, from a purely conscious, willful act to one that can come under control of the environment [15-17] – the “holy grail” of nudge-based interventions.

Puzzle Piece #2: The Theory of “Mindsets”

The second puzzle piece, then, specifies some of the psychological consequences that arise from a person being in any one of the journey phases. These consequences can be thought of as the mind’s way of doing what needs to be done to be successful at the jobs to be performed within each phase:

  • In the “predecisional” phase, good decision-making depends on a person being open to information about the various behavioral goals to consider and having a proper understanding of their likely outcomes. As such, a decision will be optimal, or close to it, when the consideration of immediate and future pros and cons is both maximally thorough and minimally biased. It’ll also be closer to optimal the more realistic one’s evaluation is of the chances of success or failure with each possible goal.
  • By contrast, the “preactional” and “actional” phases require one to be single-minded in one’s goal commitment, and to take steps that will protect behaviors in service of that goal from distractions or doubt. In that case, success will be more likely the more one allows a certain amount of bias to infect the belief that the mission is valuable and that success is likely. It’ll also be more likely if one puts certain limits on the consideration of incoming data, as anything superfluous may just serve to create new, unnecessary questions and distract from what needs to be done.

Thus, each phase pulls for a certain kind of “mindset”, or constellation of ways of thinking, that, in the predecisional phase, is open, balanced, and less prone to self-serving biases, but, in the preactional and actional phases, is more closed, imbalanced, and prone to biased justifications and risk beliefs. Note that this doesn’t mean that people won’t be biased in the predecisional phase, or that they’ll only use biases in an adaptive way in the preactional and actional phases; it only means that they’ll be less or more inclined toward certain biases depending on where they are in the journey. A few example features of these mindsets appear below [see 2-5 for details]:

Post3_Graphic2_Mindsets

What I Like About the Action Phases Model

So, in sum, the Action Phases model offers two puzzle pieces:

  • A model of intended action, aimed at achieving one or more goals, that specifies the phases a person goes through to commit to and complete that action, and provides enough information to give one a decent sense of what each phase entails
  • A second conceptual layer that further illuminates the psychology of the action phases and, thus, gives us even more information about behavioral phenomena likely to emerge in each phase

Together, these pieces have four things going for them that, I think, make the model an excellent item to have in the behavioral analysis toolkit:

#1: The practical payoffs can be immediate

For starters, consider the Rubicon model. If we were to apply this model without ever considering anything about the associated mindsets, we’d still have more than enough to begin thinking through where a behavior of interest might break down, and what kinds of interventions might make sense to consider for the behavior we hope to see. We’d also have a way of thinking about the order in which intervention components might need to appear, and the methods to be used to determine where a person is on their journey to inform intervention delivery timing. I’ve seen firsthand how, with something as basic as a static website for encouraging a particular everyday behavior, simply clustering and ordering content by the action phases and embellishing it to fit the jobs to be done at each phase can start to produce a reliable lift in engagement, consistent with what one would expect if those modifications were helping to make the content feel more behaviorally relevant. If we were to go a step further and add in the information about the mindsets, we’d then benefit from yet even more opportunities to think through the type of content to serve people, along with the amount and complexity of it, depending on the action phase we think they’re in.

Lastly, just as work on the preactional phase has given rise to the “implementation intentions” intervention, so have other features of the model inspired interventions such as “mental contrasting” [18] which have been subject to published research. Thus, the Action Phases model doesn’t only support designing interventions from scratch; it also points the way toward tested tactics that we can pull from the literature and implement within an intervention’s broader design, using the model as a guide for where and how to implement them.

#2: The Rubicon model accords with a lot of what we know about behavior

One nice feature of the Rubicon model specifically is that it is a model of phases, and not of either-or stages that one progresses through with only small amounts of backsliding. This fits nicely with the ways in which behavior can be nonlinear and context-bound: There’s simply no reason why a person cannot be in, say, the preactional or actional phases and still show some degree of thinking consistent with what occupied their minds in the predecisional phase. There’s also no reason why they couldn’t jump back and forth between phases, following whatever pattern would make sense depending on the difficulties they’ve run into in the later phases.

Yet, it's not just in this additional bit of psychological realism that the model excels. As I noted earlier, the model comes from a broader tradition of work on “self-regulation” – an area that is chock full of formal self-regulation models along with independent lines of research on goal setting, learning, feedback usage, and other behavioral processes relevant to goal-directed action. Given this heritage, it’s not surprising that the model’s contours fit much of the work within this literature, providing a set of buckets within which other behavioral science learnings can naturally be poured. And that’s important, because it sets the stage for the third reason to like the model, which is that –

#3: The Rubicon model makes an excellent foundation for custom model-building

I said at the top of this post that I liked the Action Phases model for the role it can play in building a model for a behavior we’re looking to understand and change – and this is one of the reasons for it. It follows from one of the Rubicon model’s stronger virtues, which is that it doesn’t try to specify too much. As such, it provides room for us to ask questions about specific behavioral processes that might apply at each phase, allowing us to tailor the questioning to suit our purposes, and to synthesize across far corners of the science to provide answers that make the most of the literature’s learnings. For instance:

  • In the predecisional phase, we know that people’s representations of choice options are going to be incomplete and distorted, even if there’s a pull to think more rationally. How are those representations, and the feelings people have about them, affected by context, by what’s top of mind, or by biases in what people recall about the past, or how they picture and project themselves into the future? How are their estimates of what’s likely to follow from these choice options influenced by intuition, feelings, and context? Also: People don’t just evaluate potential outcomes in a vacuum; they look at them in terms of their relevance for broader goals, which, in turn, gain their potency from their links to core needs and their place in people’s definition of themselves. How does the structure of these goal systems impact the strength of commitments to certain courses of action? What are the various contents of people’s self-concepts, and what role do they play in their thoughts, feelings, and intentions around the options?

We can also ask similar questions about the actional phase:

  • What types of feedback count in this phase? Are there circumstances in which people may rely on too much or too little feedback, and what are the consequences of either? Also, people need to be able to react to feedback in ways that are going to facilitate intended behavior and good decision-making. What methods do people use to cope and regulate their emotions so that their responses to feedback will be productive? Under which circumstances might these coping mechanisms begin to break down?

These questions cover a host of issues that are relevant to intervention development, the answers to which could very well take one on a tour of at least nine or ten different bodies of published work. The Action Phases model is, itself, silent on much of it – but that’s irrelevant. What’s important is that the model is congenial to what these lines of work can tell us about the processes our questions are designed to help us understand. By tying what we learn back to the structure of the model, we’re able to use that structure to build a comprehensive, coherent picture of behavioral drivers while capitalizing on the practical payoffs that the model already provides. And, by being given the flexibility with which processes we pull into the model, we can build a model for our case that identifies candidate drivers to explore in research, or begin targeting, that make sense for the specific problem we’re trying to solve.

#4: The model doesn't require you to buy all of it

Finally, we’ve described the overall Action Phases model as containing two puzzle pieces, one of which is the Rubicon model, the other of which contains the constellations of thinking tendencies, or "mindsets", to which the action phases presumably give rise. Now, should we “buy” the mindsets piece of the model? Maybe. I’ve seen enough within these studies to find the mindset effects interesting, but also enough to make me wonder how solid these effects really are. No matter. The mindsets are sufficiently separate from the basic propositions in the underlying Rubicon model that it’s as if this one feature of model comes with a certain kind of severability clause: Nothing about the rejection of the work on mindsets necessarily invalidates the Rubicon model specifically. And, again, that gives us the flexibility to focus on either the Rubicon model alone or the entire package, based on what we believe to be the strength of evidence for each piece and the costs or benefits of being more conservative or liberal in our decision about what to use.

Summary

I noted earlier that there’s never just one model of behavior that can be the solution to all of our behavioral insight and behavior change challenges – and the Action Phases model is no exception: It’s simply one conceptual tool among many that can be brought to bear in the analysis of behavior with which we’re concerned. Yet, I hope I’ve been able to show that it is actually a pretty powerful model to have in the toolkit, as it can provide an excellent foundational building block for pulling in a vast swath of behavioral science learnings, tying them together in a way that’s fit for purpose, and surfacing behavioral drivers that could potentially be targeted for intervention. It even comes with interventions that have been inspired by the model and subject to rigorous testing. Together, that can allow us to apply behavioral science learnings in an organized, efficient way while still allowing the applications to strive for as much depth as is possible or needed.

References (Were We Got Some of This)

  1. Gollwitzer, P.M. (1990). Action phases and mind-sets. In E.T. Higgins & R.M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 53-92). New York: Guilford.
  2. Gollwitzer, P.M., & Bayer, U. (1999). Deliberative versus implemental mindsets in the control of action. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp. 403-422). The Guilford Press.
  3. Gollwitzer, P.M. (2012). Mindset theory of action phases. In P.A. van Lange (Ed.), Theories of social psychology (pp. 526-545). Los Angeles: Sage.
  4. Keller, L., Bieleke, M., & Gollwitzer, P.M. (2019). Mindset theory of action phases and if-then planning. In K. Sassenberg & M.L.W. Vliek (Eds.), Social psychology in action (pp. 23-37). Springer, Cham. https://doi.org/10.1007/978-3-030-13788-5_2
  5. Keller, L., Gollwitzer, P.M., & Sheeran, P. (2020). Changing behavior using the model of action phases. In M.S. Hagger, L.D. Cameron, K. Hamilton, N. Hankonen, & T. Lintunen (Eds.), The handbook of behavior change (pp. 77-88). Cambridge University Press. https://doi.org/10.1017/9781108677318.006
  6. Adriaanse, M.A., Vinkers, C.D.W., De Ridder, D.T.D., Hox, J.J., & De Wit, J.B.F. (2011). Do implementation intentions help to eat a healthy diet? A systematic review and meta-analysis of the empirical evidence. Appetite, 56, 183-193. https://doi.org/10.1016/j.appet.2010.10.012
  7. Bosch, I.C., Vila, I., & Redondo, R. (2019). What makes implementation intention interventions effective for promoting healthy eating behaviours? A meta-regression. Appetite, 140, 239-247. https://doi.org/10.1016/j.appet.2019.05.024
  8. De Vet, E., Oenema, A., Sheeran, P., & Brug, J. (2009). Should implementation intention interventions be implemented in obesity prevention? The impact of if-then plans on daily physical activity in Dutch adults. International Journal of Behavioral Nutrition and Physical Activity, 6, 11. https://doi.org/10.1186/1479-5868-6-11
  9. Cooke, R., McEwan, H., & Norman, P. (2022). The effect of forming implementation intentions on alcohol consumption: A systematic review and meta-analysis. Drug and Alcohol Review, 42, 68-80. https://doi.org/10.1111/dar.13553
  10. Belanger-Gravel, A., Godin, G., & Amiereault, A. (2011). A meta-analytic review of the effect of implementation intentions on physical activity. Health Psychology Review, 7, 23-54. https://doi.org/10.1080/17437199.2011.560095
  11. Gollwitzer, P.M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in experimental social psychology (Vol. 38, pp. 69-119). Elsevier Academic Press. https://doi.org/10.1016/S0065-2601(06)38002-1
  12. McWilliams, L., Bellhouse, S., Yorke, J., Lloyd, K, & Armitage, C.J. (2019). Beyond “planning”: A meta-analysis of implementation intentions to support smoking cessation. Health Psychology, 38, 1059-1068. https://doi.org/10.1037/hea0000768
  13. Toli, A., Webb, T.L., & Hardy, G.E. (2015). Does forming implementation intentions help people with mental health problems to achieve goals? A meta-analysis of experimental studies with clinical and analogue samples. British Journal of Clinical Psychology, 55, 69-90. https://doi.org/10.1111/bjc.12086
  14. Sheeran, P. (2002). Intention-behaviour relations: A conceptual and empirical review. European Review of Social Psychology, 12, 1-36. https://doi.org/10.1080/14792772143000003
  15. Chen, X., Wang, Y., Liu, L., Cui, J., Gan, M., Shum, D.H.K., & Chan, R.C.K. (2015). The effect of implementation intention on prospective memory: A systematic and meta-analytic review. Psychiatry Research, 226, 14-22. https://doi.org/10.1016/j.psychres.2015.01.011
  16. Webb, T.L., & Sheeran, P. (2007). How do implementation intentions promote goal attainment? A test of component processes. Journal of Experimental Social Psychology, 43, 295-302. https://doi.org/10.1016/j.jesp.2006.02.001
  17. Webb, T.L. & Sheeran, P. (2008). Mechanisms of implementation intention effects: The role of goal intentions, self-efficacy, and accessibility of plan components. British Journal of Social Psychology, 47, 373-395. https://doi.org/10.1348/014466607X267010
  18. Oettingen, G., & Schworer, B. (2013). Mind wandering via mental contrasting as a tool for behavior change. Frontiers in Psychology, 4, 562. https://doi.org/10.3389/fpsyg.2013.00562

© 2023 Grey Matter Behavioral Sciences, LLC. All rights reserved.