Should we talk about controlling as a behavior?

Isn’t it obvious? We often talk about someone controlling his or her environment – especially someone else’s behavior – implying that controlling is a behavior of some sort. We also talk about control as the outcome or consequence of such behavior, perhaps a type of positive reinforcer, as when we are successful in engineering a particular change in the environment. Is there a distinct and useful operant at the root of these vernacular references? Is “controlling” a good label for a relationship between certain actions and their environmental effects?

Our colloquial approach to the question

It may seem a cop out to remind ourselves that when we talk about control as behavior or consequence we are just emitting a bit of verbal behavior, but this is the best way to start answering this kind of question. Ever since Skinner showed us the way in his 1945 paper on operationism, figuring out what a word “means” begins with looking for the sources of control over our “use” of the word. (The quotes are to remind ourselves that from a radical behavioristic perspective words don’t have meanings in the everyday sense, and we don’t use them any more than we use other behaviors.)

In common parlance, we tend to talk about controlling as a kind of behavior when we see a relationship between someone’s actions and certain changes in the environment, especially the behavior of someone else. We might watch a parent get a child to comply with a request or a friend get others to help move furniture to a new apartment. We tend to summarize situations involving a person doing something that changes their environment in a reinforcing way with references to control as both action and result.

We usually have more specific vocabulary for describing the particular actions that are said to be controlling and for describing the environmental changes that result. For example, we may observe that our friend promises pizza and beer to get help moving the furniture, which then winds up in a new apartment. Whatever the specifics in one situation or another, references to controlling seem to share a certain kind of relationship between some behavior and its consequences. We seem unlikely to label any particular actions as controlling if we don’t see the appropriate consequences – usually reinforcing outcomes. We are also less likely to talk about control if we only see certain changes in the environment but not the behavior that may have led to them. In other words, both sides of the action/outcome relationship seem to be the key to applying the label.

Does this analysis of colloquial verbal practices suggest that we should label behavior that produces reinforcing changes in the environment as controlling? Of course, any behavior that tends to produce reinforcing outcomes has a regular place in our repertoire, so in this sense much of our behavior would fit such a definition – in fact, too much. The range of actions that would qualify as controlling would be both functionally and topographically inexhaustible and therefore perhaps not a useful way to define an operant.

We could reduce this breadth by requiring awareness on the part of the person doing the controlling. After all, we may be more likely to “use” the word when we see malice aforethought, so to speak. In everyday lingo, we often refer to someone engaging in this sort of behavior when it is clear that the person is aware of his or her actions and able to describe the reinforcing outcome in advance. More technically, this qualification means that the person is engaging in verbal behavior – though perhaps covert – about their efforts and the desired outcome. This approach comes dangerously close to implying intent, however, carrying teleological overtones. Although adding this requirement to the definition of controlling as an operant would often fit our use of the term, there are certainly many circumstances in which we would still want to label behavior as controlling even though the person is presumably not able to articulate the relationship between their actions and a reinforcing outcome.

A caveat or two

So where does this brief review of possible influences on the way we talk about controlling behavior lead us? We talk this way when we observe a (usually) positively reinforcing relationship between some behavior and certain outcomes, often arranged with awareness. These actions are typically directly responsible for the resulting consequences, but we cannot circumscribe the “controlling” behavior or its outcomes in terms of topography because the possibilities are limitless. This is not so simple as defining the behavior of making a cup of tea or driving a car.

Before going further, however, a caveat or two may be in order. Our analysis of influences on everyday dialect is not about whether there is such a behavior as controlling or a consequence we might call “control.” Behavior is the continuous interface between organism and environment. It doesn’t come prepackaged into separate behaviors. We merely overlay certain distinctions that serve our purposes. This is often a challenging exercise, and the results need not neatly reflect everyday vocabulary. As behavior analysts, we are looking for distinctions that are consistent with our science and its philosophy, as well as useful in our research and service agendas. The question is not whether controlling is a behavior but whether applying this label to a certain kind of relationship between particular actions and their consequences is a useful thing to do.

How we define and label target behaviors is critical to the value of our applied research literature. One of the reasons our literature is useful is that the way we define target behaviors minimizes subjective colloquial influences in favor of a scientific understanding about how behavior actually works. When we review a group of studies that focus on a particular target behavior – aggressive behavior, noncompliance, working math problems, and so forth – we assume that their findings have a collective value that is greater than that offered by any one of the studies alone. This combined value partly depends on a shared focus on the same behavioral features, as implied by the common label for the target behavior. If we are looking at what the literature says about interventions for reducing aggressive behavior, for instance, we assume that the included studies addressed similar aspects of this kind of behavior and therefore offer interdependent findings that stand a good chance of generalizing to our interest in a particular case of aggressive behavior. If these studies targeted behaviors that did not share key features, such generality would be less likely.

The challenge in applying the label “controlling” to particular features of behavior is avoiding capturing aspects that are too broad or varied from one application to another. That is, researchers and practitioners might use the label in so many different ways that the accumulating literature said to be about the behavior of controlling would not have useful generality. Each study might be methodologically sound, but the referenced literature might not yield coherent findings that help us understand such behavior. What one study shows about “controlling behavior” might not tie into what is revealed by another study supposedly addressing the same behavior. Such a literature might grow large over time but fail to advance the science or its technology.

This challenge is shared by many of the labels we use to identify particular behaviors. In fact, we might think of our labels for particular response classes as spanning a continuum of topographical or functional specificity. At one end our references may be pretty detailed, such as “checking our Facebook page” or “carrying groceries into the house.” At the other end, however, they may be quite broad, such as “helping” or “socializing” or ‘”sharing.” In all cases, the question we should ask is whether our references to supposedly distinct behaviors are helpful to our science and practice. (See Labeling behavior, posted on September 18, 2016, for further discussion of this point.)

Skinner, 1935

In wrestling with these issues, it might be useful to remember that Skinner suggested some rules for how we define response classes, though he would no doubt argue that all he did is figure out how nature does the job. In a paper published in 1935 – three years before he published The behavior of organisms – he resolved a long-standing problem in psychology. He proposed that the “natural lines of fracture” that reveal distinct, organized pieces of behavior could be discovered by identifying functional relations between classes of responses and classes of consequent stimuli. (See Chapter 4 in Johnston & Pennypacker, 2009 for a more detailed discussion.) The resulting behaviors might be small and encompass narrowly identified forms of responses and stimuli, or they might be large and incorporate topographically varied responses and stimuli.

The behavior we informally identify as controlling obviously involves the latter circumstance. The topography of what we might do to control someone else’s behavior or other environmental events differs greatly from one situation to another, as do the resulting changes in the environment that influence the likelihood of further controlling behavior. What would justify a reference to controlling as a coherent response class or behavior is that these seemingly different responses and their consequences are functionally related – that they depend on each other in the same way. Controlling responses must serve the function of producing certain stimulus outcomes, which must serve the function of changing the likelihood of future responses of the same sort.

This approach to defining controlling as an operant is doable, but few are likely to go to the trouble of empirically demonstrating such a functional relation in the course of conducting a research project or practical intervention focusing on “controlling behavior.” After all, such demonstrations are uncommon in the applied literature for much more topographically constrained target behaviors. Can we find a way to apply this vernacular label so that it captures a meaningful functional relationship that can be reliably identified or would it be better to avoid technical references to controlling as a behavior? This question may also be raised for other broad everyday references to behaviors such as sharing, cooperating, helping, playing, and so forth. These examples of expansive behavioral labels should remind us that just because our colloquial dialect supplies us with options for labeling large chunks of behavior does not mean we should cobble together a technical way of accommodating them. Will references to a behavior called “controlling” help researchers understand such behavior? Will it help practitioners meet their client’s needs?

A recent post concerning labeling (Johnston, 2016) recommends a conservative approach. Although there is no particular limit to the “size” of an operant, too much breadth raises the risks already mentioned. These risks can be minimized by 1) focusing on relatively narrow and specific actions and consequences, 2) clearly defining target behaviors in functional terms, and 3) labeling the resulting behavior in ways that minimize wide-ranging interpretations. If a researcher or practitioner is interested in what might be commonly called “controlling behavior,” it might be most useful to the ABA community to focus on the particular features of responding set up by an experimental preparation or characterized by a client’s actions. These features, such as “giving other people directions” or “promising to provide reinforcers,” may allow more precise definition and be easier to measure than more broadly described activities. In other words, it may be best to leave colloquial terms to everyday dialect and take a more scientific or technical approach to how we define and label target behaviors.

References

Johnston, J. M. (2016). Labeling behavior, September 18.

Johnston J. M. & Pennypacker, H. S. (2008). Strategies and tactics of behavioral research, Third edition. New York: Routledge Press.

Skinner, B. F. (1935). The generic nature of the concepts of stimulus and response. Journal of General Psychology, 12, 40-65.