Think first, but act then only if you want to: how cognitive defusion works?

Cognitive defusion is one of the most researched components of the psychological flexibility model underlying Acceptance and Commitment Therapy, but yet remains one of the most intensely debated in terms of whether it refers to a set of techniques, a process, or an outcome. I am delighted to present Daniel Assaz, a clinical psychologist and doctoral research student at the University of Sao Paolo in Brazil, who published one of my favorite articles of 2018 (in The Psychological Record) that aimed at unpacking what the basic processes of change are in cognitive defusion. I keep a close eye on research from Brazil because I believe that some of the most currently innovative and exciting behavior analytic work is coming out of Brazil. Enjoy Daniel’s wonderfully insightful and nuanced blog. Dr. Ian Tyndall, University of Chichester, UK

“If I were to really believe in everything I think, I would go crazy”

Mário Quintana, Brazilian poet

As a behavior analyst, one of the things that I have to constantly remind myself of is that scientific concepts do not reflect “the real world”, but rather are terms that have been constructed because they allow us to interact with the environment around us (including other people) in a more effective manner. That is, they are useful. Take, for instance, the picture below. What is it?

Most people would say it is “a bee”. By naming it “bee”, we can better interact with other people and influence their behavior, for example, helping a friend protect his beverage by saying “look out, there’s a bee near your coffee”; or regulate our own behavior, keeping our distance from it (because bees have stings) but not necessarily leaving the room, unless you are allergic to it (because bees are not aggressive). To sum it up, naming it “bee” helps us influence the behaviors of others and our own in a useful way.

But you wouldn’t be wrong if the answer you gave for my question was “an animal”, “an insect” or even “an Apis mellifera”. All of these terms are adequate answers. The difference between all four of them is their degree of precision. The picture really does show an animal, but so would a picture of an elephant, a salmon or your pet. Thus, calling it “an animal” has a low degree of precision. Alternatively, naming it “Apis mellifera” narrows it down to the species scientific name and, thus, has a high degree of precision, distinguishing it from other species of bees.

What is interesting to notice is that the varying degrees of language precision outlined above are more or less useful depending on the context. When a young child is expanding his vocabulary, saying it is “an insect” can be an important developmental step in teaching him to discriminate between different types of animals. When our friend is drinking coffee, saying it is “a bee” will help him look to the air and not the ground in order to protect his drink, as he would do if the threat was “an ant”; but he does not to know what species of bee is it in order to do so. And when a biologist is conducting an ecological evaluation of an ecosystem it is important for him to discriminate between Apis mellifera and other species of bees. Therefore, it is crucial to be able to flexibly alternate between different levels of language precision, because one will be more useful than the other in specific contexts.

All of this is probably very intuitive, but when we take a closer look at behavioral science, the capacity of flexibly alternating between different levels of language precision is currently at the center of a heated debate. As a scientific community, Behavior Analysis has historically favored the use of concepts with high precision, such as “positive reinforcement” and “simple discrimination”, arguing that such precision would lead to better prediction and control of behavior. That is, that they are more useful to behavioral analysts. But, recently, other behavioral scientists, aligned with the scientific strategy of Contextual Behavioral Science (CBS), have been arguing in favor of adopting concepts with lower precision, that would be more useful in the context of applying and disseminating behavioral principles: middle-level terms.

Acceptance and Commitment Therapy (ACT) is a contextual behavior therapy that has been at the center of this debate due to the fact that its model is most commonly presented through six middle-level terms: acceptance, cognitive defusion, contact with the present moment, self-as-context, values and committed action. Despite the strengths of these terms regarding dissemination to wider audiences, not linking them to basic processes, with higher levels of language precision, can hinder their integration with functional-analytic reasoning, which is at the core of behavior-analytic philosophy and theory. But, if such link is explicitly laid out, the speaker is able to flexibly alternate between different levels of language precision to communicate to a lay audience, create solid scientific lines of research and integrate ACT concepts within a functional-analytic case conceptualization tailored to each client.

For the last few years, our laboratory at the University of São Paulo, in collaboration with external researchers, has been investigating how behavior analysts have been using the term “cognitive defusion” in order to evaluate the existence of the link between this middle-level term and basic behavioral process and, thus, allow for flexible use of concepts, enhancing their usefulness.

Surveying the ACT literature, what we initially found was much conceptual ambiguity in how behavioral scientists were using the term “cognitive defusion”. At times, it referred to specific procedures employed by therapists. Occasionally, it described the behavioral process by which such procedures worked. And here and there, it was used to specify the clinically relevant outcomes produced by these processes. Sometimes all three uses of the term were seen mixed together in the same paragraph. All in all, it was a clear case of conceptual ambiguity, in which one concept refers to three distinct type of phenomenon. At the end of the day, stating that cognitive defusion procedures work through cognitive defusion processes to promote cognitive defusion is not very useful. Can you imagine a doctor explaining to his patient that he will use chemotherapy procedures that promote chemotherapeutic processes leading to chemotherapeutic results? It would be very confusing, to say the least.

In order to enhance conceptual clarity, we began by separating procedures, processes and outcomes related to cognitive defusion, as stated in the literature. First, we outlined that the desired outcome is to reduce the verbal control of the client’s thoughts over his behavior. Clinically speaking, that is especially important when such verbal control is promoting response patterns that are rigid and move him away from his values, named fusion in ACT terms. In other words, cognitive defusion does not aim to suppress thoughts or change their content, but reduce their capacity to regulate our behavior: to think but not necessarily to act according to what we think, creating a gap to choose what we really want to do in the moment.

This psychological experience resembles the stance we learn to develop with dreams. When you were a kid and had a really bad dream, waking up still in fright, what would your parents often say? Probably something along the line of “it was only a dream”. This sentence does not deny that the experience was real or uncomfortable, only that we shouldn’t take them so seriously: do not argue with your partner if he or she cheated on you in your dream. Cognitive defusion aims to build a similar experience with our own thoughts: just because you are thinking that people will judge if you express yourself, it doesn’t mean that you shouldn’t do it anyway if it is important to you.

Moving away from the desired outcome, we identified a wide array of procedures related to cognitive defusion. To some people they may appear as highly creative or extremely weird. Either way, evidence is accumulating that at least part of them can be effective. To summarize, six categories of procedures were outlined:

  • Playing with thoughts through repeating them for 30 seconds, singing them in a familiar tune or pronouncing them really slowly.
  • Objectifying thoughts, attributing physical properties to them.
  • Observing the stream of thoughts, often aided by objectifying them, such as leaves flowing down a stream.
  • Disrupting the link between thought and action, creating and drawing attention to experiences of incongruence between the thought’s content and the person’s actions, when he plainly disobeys them.
  • Identifying components of thinking, naming them bluntly as thoughts or evaluations, for instance.
  • Questioning the process by which people construct their personal narratives and the reasons for their behavior.

But the part we were most interested in was how the literature conceptualized what was going on, behaviorally speaking, when clients went through these types of procedures, what were the behavioral processes involved. After all, if we understand how these procedures work, we are in a better position to optimize their effectiveness in clinical practice; to make the therapist’s creativity and personal style flourish in designing new and exciting procedures; and to disseminate this knowledge into therapists’ trainings.

The most widely held account for the process involved in cognitive defusion was that thinking is able to control behavior under specific stylistic and structural parameters of speech. These parameters were called the context of literality. Thus, removing such parameters would reduce the control of thinking over behavior. However, this definition is based on the assumption that because cognitive fusion and defusion are opposite regarding outcome, they should also be opposite regarding processes. However, this is not necessarily true. Human behavior is complex, to say the least, and there isn’t always symmetry between outcome and process. This means that if you see a child being afraid of a dog, you can’t tell just from this scene if this behavior was learned because he has been previously bitten by a dog, seen someone be bitten by a dog, or instructed that dogs are dangerous animals. Similarly, even if the learning process was verbally acquired (that is, the child was told that dogs are dangerous), the aversive functions of dogs can be reduced through multiple process, including direct exposure to dogs or new instructions on how friendly dogs are.

In order to account for such complexity in everyday human behavior, the conceptual proposal of our laboratory is that there are different pathways through which the goal of reducing the control of thinking over behavior can be achieved. We differentiated three main processes that underlie the variety of procedures found in the literature. To illustrate them, pause for a second and identify a thought that is difficult for you and sometimes prevents you from doing things that are important.

The first pathway is to redirect the client’s attention from the symbolic meaning of thoughts to their physical properties, changing stimulus control. Take the thought you identified earlier and try to say it out loud, but very slowly, clearly emphasizing every syllable. Do this a couple of times. Chances are your attention drifted somewhat from its meaning to how the words sound or how your mouth, tongue or throat moves while speaking. Or take the thought and attribute physical properties to it. If the thought is regarding the pressure you feel to do something, you can imagine it as an old, heavy, gray anvil that weighs down on your back; or as an old cuckoo clock constantly looking at you from across the room, going tic-tac, tic-tac. Observe its shape, its color, its size, its weigh, its texture etc.

If these exercises were successful, the experience should be similar to when you are travelling abroad and trying to read a different alphabet but have never learned it (such as Russian, Chinese or Arabic for English speakers): you can only react to the physical properties of the symbols, but not their meaning. Or to when you are singing a song but focused on the melody and not the lyrics, which can even catch you off guard when you stop and pay attention to what the song is actually saying. If the exercises were not successful, don’t worry. We only spent a minute doing them and some thoughts have such a grip on us that it will take some time and some practice to let go of them.

The second pathway is to draw the client’s attention to previous experiences or create new ones in which he disobeys his thought’s contents, acting in a manner that is incongruent with them. Such “disobedience” responses can thus be reinforced in session when the therapist helps the client to link them with his objectives and values, promoting a simple discrimination: when you have a thought that is not helpful, disobey it. For a minute, take your mind to the thought you identified earlier. It probably prevented you from doing things you deemed valuable. But have you ever had that thought and still done what was important to you? Even if it was just once, capture that moment. How did it feel? How are you feeling right now as you remember this moment? Can you experience that thought and still do what you choose to do? This experience is similar to when you have daydreams or fantasies that are not possible or socially acceptable and, thus, you do not act upon. In the end, this means that there does not need to be a correspondence between what we think and what we do.

The third pathway involves realizing that thoughts do not occur in a vacuum. Rather, they are the products of an individual, who is behaving the way he is due to his history and the effects that specific thoughts have on his behavior. To illustrate this, take a worry you had recently. Just because you are worrying about something bad that may happen in the future, that does not mean this event is likely to happen. It probably means that this event is important, that you have had experiences with bad things happening to important parts of your life in the past; and that worrying about the worst case scenario can help you come up with creative solutions on how to prevent it from happening. Even if such thoughts can seem irrational or cruel, they might be the best ways we are able to respond to the situation at the time.

Once you are able to see this historically and functionally-driven speaker which is yourself, you are able to engage with him and the worry narrative he has created in dialogue. That can be done by recognizing and appreciating his intentions: it’s trying to protect you from harm. Or by using humor and irreverence towards the narrative’s coherence or creativity: the worry is very compelling, but it probably left out some bad things that can also happen, it is lacking more inspiration. This experience is similar to the one we experience when a salesperson makes a claim about his product that doesn’t seem reliable: we are readily skeptic, with reason, that he is only saying that in order to sell, no matter what.  Or to the one of a movie critic, analyzing the dialogues and the script and pointing out its good and bad aspects. In the end, what we are doing is creating a conditional discrimination that includes the thinker, the thought and how we respond to both of them. If the speaker was different, our response would change accordingly.

To sum up, we have reached a proposal for conceptual clarity regarding cognitive defusion in which it is best understood as an outcome, because defusion-related procedures and processes vary a great deal, while outcome is a unified phenomenon: the reduction of the verbal of thinking over behavior if such control is evoking rigid and problematic responses. And such outcome can be achieved through different processes that underlie a wide variety of procedures the clinician can choose from or create anew: change in stimulus control, simple discrimination and conditional discrimination. Future research is encouraged to evaluate how this theoretical conceptualization fits empirical data and, if it does not hold up, change it towards a better, more useful one. But even if that happens, the work outlined here of separating between outcomes, procedures and processes and attempting to operationalize them is an important step in the same direction.