Let’s just lay it out there: There are only two fundamental approaches to defining a behavior (response class). This limitation stems from the nature of behavior as a physical phenomenon. The definition of behavior in Strategies and tactics of behavioral research (Johnston & Pennypacker, 2009) goes as follows:
Behavior is that portion of an organism’s interaction with its environment that involves movement of some part of the organism (p. 31).
It’s pretty obvious that movement is an unavoidable characteristic of behavior. It may be less obvious that the reference to “interaction with its environment” adds the possibility that these movements may have effects on the environment, which may in turn have effects on behavior. Our science has established the intricate details of these reciprocal effects, which we describe in terms of functional relations.
Topographical and functional definitions
Any particular behavior may be defined in terms of these two characteristics. A behavior may be defined on the basis of the form or topography of its responses as movements in three-dimensional space. Alternatively, the same behavior may be defined in terms of the functional relations these responses may have with the environment. Each type of definition ignores the criteria for the other, although a definition may be put together that uses both kinds of criteria. Chapter 4 in S&T covers in detail the topic of selecting and defining response classes.
The characteristics of behavior as a physical phenomenon allow only these two fundamental types of definitions. Although you can overlay either type with variations and refinements of all sorts, they always reduce to either a topographical or a functional core. Let this sink a bit in before reading further.
Operational definitions 1
At this point, some readers are probably wondering about operational definitions. Where do they come in? You may be surprised at the answer.
References to operational definitions are routine these days in ABA literature and textbooks (e.g., Cooper, Heron, & Heward, 2007). The general idea is to come up with a definition of a target behavior that will guide measurement decisions. A mere label for the behavior rarely provides sufficient direction, so the researcher or clinician describes the features observers are supposed to use to decide which of all the things an individual does should be recorded as an instance of the target behavior and, by exclusion, which should not. For example, an intervention may target “tantrumming,” but observers need to know exactly what about this kind of behavior should be recorded or not.
What seems to justify calling the resulting definition “operational” is that it specifies the observing criteria that are supposed to be used in a particular project or clinical intervention. The same summary label for the behavior (e.g., tantrumming) might be used in many other circumstances, but it is “operationalized” in each case by a particular definition. An operational definition specifies at least part of the operations (observing and recording) involved in measuring a target behavior. This reasoning makes sense, of course. A simple label for a behavior is almost always an inadequate basis by itself for guiding observing and recording procedures. There will typically be good reasons for why a detailed definition of any one behavioral label will need to vary from one case to another.
But when is a definition of a target behavior not operational? Can you have one definition of a particular behavior that is operational and another definition of the same behavior that is not operational? Answering this question requires a clear understanding of what it is we are defining. We are not defining the label we are using to identify the behavior, but the actual behavior itself. Given its features, what about it do we want to record? Any behavior will have far more characteristics than we are interested in, and a definition specifies which ones we want to capture and which ones we want to ignore. Our choices may influence how we decide to label it, but we cannot escape making these decisions.
So, it is impossible to define a behavior for purposes of measurement in a way that does not require these decisions about the features that are or are not of interest. That is, all definitions are operational. We cannot usefully talk about a particular behavior without some clarity regarding the key features that describe our interest. If we’re talking about tantrumming, we need to know about the key topographical and functional elements that warrant that label. We always have to talk in operational terms because behavior is complex and varied, and summary labels just don’t give us the information we need for practical (effective) discussion.
This argument means that there is no reason to specify definitions of behavior as being operational in nature. They are all operational, and the adjective is therefore unnecessary and useless. This point is not merely a matter of terminological niceties. The task of developing operational definitions may distract us from thinking about details of topography and function that are important for different reasons in different circumstances. This is a serious side effect. Crafting the right definition of a target behavior in these unavoidable dimensions is critical because different definitions of the same behavior are likely to reveal different effects of an intervention. We always have to ask how the data might be different if we had defined the behavior differently.
Operational definitions 2
The sense of the phrase “operational definition” had another life before this more recent terminological thievery came along, and it still does. There was a time when talking about operational definition involved a discussion of mentalism, operationism, and Skinner’s famous 1945 paper. As explained in Box 10.4 on operationism, and in more detail by Jay Moore (1980, 2008), the principle of operationism came from a physicist named P. W. Bridgeman. He argued that “we mean by any concept nothing more than a set of operations: the concept is synonymous with the corresponding set of operations” (Bridgeman, 1927, 5). (Also see Box 4.5 in S&T.)
Psychology found this approach useful because it offered a way to deal with mental events without worrying too much about their existence. One thing led to another, and what evolved was the view that language is a symbolic activity in which private subjective meanings have a mentalistic existence. That is, words have meaning in a private mental lexicon.
Not surprisingly, Skinner had serious problems with this approach to operationism, and he delivered a paper explaining his contrary views at a conference on operationism (Skinner, 1945). He proposed that the meaning of a word lies not in a mental dictionary, but in an analysis of the environmental factors that determine the emission of the word as a verbal response. In other words, he did not treat verbal behavior any differently from other behavior. The contrast with psychology’s approach to operationism was stark.
Skinner’s 1945 paper is famous, at least among behavior analysts, because his argument was immediately accepted as an important element in the conceptual foundation of the field. As a result, the phrase “operational definition” was widely understood in the context of Skinner’s paper as referring to tactic for understanding everyday terms through a search for their sources of control as verbal responses. In other words, the only occasion for behavior analysts to talk about operational definitions was when the task was understanding the “meaning” of term referencing personal qualities that were at least unclear and often fictional. For behavior analysts, the phrase referred to operationism as a topic or, perhaps, to the tactic Skinner gave us to understanding the “meaning” of verbal behavior in a way that avoided mentalism.
We are still young
There was a time when this is all that “operational definitions” meant. Why the past tense? Because many of today’s ABA practitioners refer to all definitions of target behaviors as operational definitions because that is what their textbooks told them to do. It doesn’t take too long before everyone’s doing it, and many terminological issues have crept into the field’s technical vocabulary this way. Terminological slippage is hardly unique to behavior analysis, of course. The struggle to clean up technical terminology is an ongoing battlefront in all scientific disciplines. A field’s terminology reflects its understanding of its subject matter and, thus, its maturity. In some ways, we are still young.
Although ABA marks its beginnings in the 1960s, the vast majority of today’s practitioners were trained relatively recently. The BACB certification program only began a few years ago, and not that many of today’s 20,000 certificants were practicing before its inception. The BACB credentialing program established contingencies that not only encouraged training programs to blossom and attracted students to them, but also created a growing market for textbooks. For most of today’s practitioners, these texts define ABA, never mind what has gone before.
Another reason for the conflicting references to operational definitions lies with the fact that the BACB’s curricula standards are constrained by its limitations as a credentialing body. The BACB requirements for approved course sequences do not (yet) include meaningful training in the philosophical foundations of the field. It cannot simply force this topic on programs, however, because the task standards that drive curriculum requirements are determined by practitioners themselves, rather than by a committee or task force of some sort. There are good reasons for this approach, which is a policy to which all credentialing bodies must adhere. Nevertheless, one of the side effects for our young field is that the importance of its philosophy of science – radical behaviorism – has yet to reach the level of appreciation needed to assure inclusion in mandated training curricula.
Together, these influences mean that many ABA practitioners not only view all definitions as operational, but are likely to fail to be aware of, let alone appreciate, the importance of Skinner’s view of operationism. This post explains the first problem and at least identifies the second. The larger problem is that our field faces a variety of other terminological issues as well. What are we going to do about them?
Bridgman, P. (1927). The logic of modern physics. New York: MacMillan.
Cooper, J., Heron, T.E., & Heward, W.L. (2007). Applied behavior analysis (2nd ed.) Upper Saddle River, NJ: Pearson.
Johnston, J. M. & Pennypacker, H. S. (2009). Strategies and tactics of behavioral research (3rd edition). NY: Routledge..
Moore, J. (1980). On the principle of operationism in the science of behavior. Behaviorism, 3, 120-138.
Moore, J. (2008). Conceptual foundations of radical behaviorism. Cornwall-on-Hudson, NY: Sloan Publishing.
Skinner, B. F. (1945). The operational analysis of psychological terms. Psychological Review, 52, 270-277.