Reading 2.4 – Identifying a behavior

Chapter 2 – It’s Just Verbal Behavior

Reading 2.4 – Identifying a behavior

A recent thread on the Applied Behavior Analysis Facebook page asked the following question: “Can non-verbal animals mand or is this strictly a human verbal operant?” Perhaps not surprisingly, most of the resulting reactions suggested a variety of behaviors owners considered mands in their pet’s repertoire. These examples generally took the form of an animal doing something that resulted in the owner responding in some presumably reinforcing way.

Begging a better question

I’m not going to propose a direct answer to the presenting question, if only because a thorough discussion would require more than a page or two of consideration. Instead, I’d like to raise a more general issue that is unavoidably involved. How do we know when we are correctly identifying a particular behavior and distinguishing it from others? After all, behavior is an unceasing flow of organism-environment interactions. There is no parts manual that tells us how to subdivide this intact phenomenon into useful pieces in a non-arbitrary way, much less explaining how the parts function as a whole. We depend on our science to supply these details, but the “rules” about identifying and distinguishing operants (response classes) leave much room for debate about the particulars in each case.
This issue is central to many topics and has come up in a number of these readings: “What is “___________?”” (Reading 4.2), “Can dogs feel jealousy?” (Reading 4.5), “The problem with problem behavior as communication” (Reading 2.3), “Form, function, and operational definitions” (Reading 10.4), “Isn’t this just bribery?” (Reading 6.4), “Labeling behavior” (Reading M3), and “Should we talk about controlling as a behavior?” (Reading 6.3). The varied themes of these posts suggest that practitioners often run into the challenge of deciding when it is useful to separate and label distinctive features of the continuous flow of an individual’s behavior. After all, there are truly endless possible discriminations that might be made and many temptations for applying everyday labels to particular collections of responses.

The rules

The rules about how to identify a behavior and distinguish it from other behaviors are Nature’s, but B. F. Skinner gave us a good start at figuring them out. In his 1935 paper, “The generic nature of the concepts of stimulus and response,” he explained that individual responses sharing the same functional relations with surrounding environmental events form natural classes or operants (Skinner, 1935). The resulting behaviors are termed functional response classes because they are defined by shared functional relations between similar actions and related environmental events. This approach to distinguishing among behaviors remains the foundation of our understanding of operant behavior, both conceptually and practically, though matters can certainly get more complicated than this.
Consider the behavior of opening doors as an example of this tactic. That behavior is defined by the functional relations between (1) the stimuli associated with doors and their opening mechanisms, (2) our actions with regard to those stimuli, and (3) the outcomes involving doors sufficiently opened for something or someone to pass through, presumably a reinforcing state of affairs. The breadth or focus of the definition of door opening can be adjusted by changing the criterion for the antecedent or consequent stimulus classes, which will result in somewhat different though still functional definitions of door opening behavior.

Although this example may be easy, there is no need to change this approach when the behaviors of interest seem less discrete or obvious. If we want to define a behavior we might call “whining,” for instance, it may be more challenging to identify the three-term contingency, but the exercise is the same. In general, the antecedent events might include aversive conditions of some sort, the actions would involve verbal responses in a particular tone of voice, and the consequences would be certain reactions by listeners. Of course, the necessary details would depend on different individuals and environmental circumstances.

The key to applying this definitional tactic is to be clear about the difference between a definition that captures a naturally occurring response class and one that merely reflects our interests and thereby falls short of this functional standard. In other words, the issue is whether the responses in an identified and labeled operant belong together because of shared environmental influences or their grouping reflects an a priori conviction that the labeled class ought to be a coherent behavior? Does the definition capture a natural response class or is it arbitrary, thereby violating the concept of an operant?
What about the possibility of a behavior we could call “being mean?” We might say, for example, that a boy is being mean to his little sister. Is being mean an operant? Although we have learned to label behavior this way under certain circumstances, does the fact that our verbal community taught us to say this guarantee that there is a functional response class that warrants this label? After all, whatever the boy was doing to his sister could probably be described or labeled in other ways. If he was saying unkind things to her, we could also label that as “criticizing.” Do the vernacular labels “criticizing” and “being mean” reference different behaviors? Although we might apply different labels to the same operant, the availability of a long list of everyday descriptions for aspects of behavior raises the question of whether each label justifies the search for a unique operant.
This example suggests another standard for identifying a particular operant in the ongoing stream of behavior: whether it is useful to researchers or practitioners in understanding behavior. The number of behaviors that might be distinguished using this functional rule is truly infinite, but only a subset is likely to be informative. Skinner’s distinctions among verbal operants have obviously proven to be valuable in helping us understand this otherwise overwhelmingly complex aspect of the human repertoire. In contrast, it is unlikely that searching for distinct operants uniquely associated with each of the thousands of English language labels for moods, attitudes, and traits would be equally productive. Remember that the whole point of identifying functional response classes is to improve our understanding of behavior and how to manage it.

Do animals do it?

Being clear about our tactics for identifying and labeling operants is especially important when we ask whether particular behaviors observed in non-human animals are the same as found in humans because the temptation to anthropomorphize can be overwhelming. The problem is that we often wonder about cross-species behavioral similarities from a colloquial rather than behavior analytic framework. Our vocabulary tends toward the vernacular and is chock full of mentalistic allusions.

For example, it is all too easy to ignore the unavoidable differences in behavioral histories of dogs and their owners. The most obvious difference is that a dog’s repertoire does not involve the influence of a verbal community of the sort benefiting its owner. The implications of this difference alone are pervasive and suggest many important differences between dogs and people.

So, what does it mean to propose that a particular set of actions observed in one species is the same behavior observed in a different species? Although the form of the behavior in question will often vary between different species, their functions could certainly be the same. In the case of dogs and humans, for example, is it useful to view the dog’s barking behavior as functionally the same as the human’s verbal behavior just because both involve sounds generated by a vocal apparatus? Questions such as this require examining whether the antecedent and consequent stimulus classes to which barking and speaking are functionally related are the same. If these stimulus classes – as well as the form of the behavior – differ, perhaps it would be useful to view barking and speaking as different behaviors.

To reiterate an earlier point, notice that this question is not about whether barking and speaking is the same behavior but whether it is useful to merely consider this possibility. Subdividing the phenomenon of behavior into different response classes or behaviors is something we do for our own convenience. Although we can do this in a non-arbitrary way, in the long run what counts is whether our decisions prove beneficial in our effort to better understand behavior. Will it eventually turn out to be more beneficial to view barking and speaking as the same operant or as different operants?

So, about the manding question

It is important to bring all of the above issues to the question of whether non-verbal species learn a behavior we call manding. Of course, we already have a pretty good understanding of this behavior in humans that have acquired a verbal repertoire, so the question is whether other species can acquire the same behavior. It may be tempting to focus on what could be superficial similarities, perhaps guided by our everyday dialect, but the standard for judging whether an operant in another species as the same as manding in humans is clear: cross-species operants are the same only if the components in the three-term contingencies are the same. Short of this, it might be best to proceed parsimoniously by studying similarities and differences. We should beware of allowing a common label to imply that they are the same.

In making this assessment in the case of manding, it is important to maintain a rather technical view of verbal behavior. Another post titled, “The problem with problem behavior as communication” (Reading 2.3) highlighted the risks of slipping into everyday dialect. If it seems that a dog is “communicating” by emitting a behavior that produces reinforcement mediated by its owner, for example, we might be tempted to call it manding because of certain apparent similarities to manding in humans, but that conclusion skips over the kind of technical issues reviewed here that should be considered.

In general, seeing non-human species as apparently engaging in otherwise human behavior runs perilously close to anthropomorphic convenience, even when we’re trying to be careful. There are lots of differences in biology and environmental history among species, even those in the same biological class (e.g., mammals). Although discovery of similarities among phenomena is the zenith of scientific achievement, it cannot be realized at the cost of ignoring differences.


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