Minimally Aversive Contingency Management
James O’Heare, DLBC
Companion Animal Sciences Institute
This paper provides guidance to behaviorologists, behavior analysts, and animal behavior technologists on utilizing the least aversive methods possible when constructing and implementing contingency management plans for clients and their companion animals. The strategy presented here emphasizes diligence in finding added reinforcement-emphasized approaches to resolving problematic behavior. You can find the most current version of this strategy online through the Association of Animal Behavior Professionals, linked to throughout the Professional Practices Guidelines at associationofanimalbehaviorprofessionals.com
Avoiding extremism and dogmatism
It is important to avoid exaggeration or excessive simplicity in this analysis. Accepting extreme arguments, such as that “all forms of aversive stimulation are always sure to cause irreparable harm” or that “aversive stimulation is necessary to succeed in training,” leads to dogmatic positions, which are best avoided in favor of careful consideration of the circumstances and dedication to utilize the least aversive methods possible.
An aversive stimulus is any event that functions (a) to evoke behavior that has reduced or terminated it in the past, (b) as a punisher if presented immediately following a behavior, or (c) as a reinforcer when withdrawn immediately after a behavior (Cooper et al., 2007). Added (i.e., positive) punishment, subtracted (i.e., negative) punishment, subtracted (i.e., negative) reinforcement, and extinction, even in their mildest forms, all involve aversive stimulation to some degree. Only added (i.e., positive) reinforcement involves no aversive stimulation at all. If a stimulus meets the criteria listed above, then the stimulus is classified as aversive. It is usually easy to predict which stimuli will function as aversive stimuli. Indeed, predicting which stimuli will be aversive is generally just as reliable as predicting which stimuli will function effectively as reinforcers. These predictions can be confirmed once you have implemented the chosen procedure with the stimulus in question.
It should also be noted that aversive stimulation is ubiquitous. We face hundreds, if not thousands, of aversive contingencies on a daily basis. Aversive stimulation plays a major role in controlling our daily social and nonsocial behavior. We put on seat belts to escape the buzzing sound (and tickets, and injuries). We put the back foot forward when walking to avoid falling forward. We answer a question asked of us to avoid the aversive reaction that would result from ignoring the question. We do the dishes or take out garbage to avoid spousal nagging or a lack of clean dishes. We cross the street to avoid walking past a scary looking person. We turn the wheel when driving through a curve to avoid driving into a ditch. We “nudge” a vending machine that does not quite release our product. We escape unconditioned aversive stimuli in daily life on a minute-by minute basis and come to escape conditioned aversive stimuli as well (what we call “avoidance” of the unconditioned stimulus). Most of these do not elicit strong emotional arousal or pain and the fact that a stimulus is aversive, per se, does not mean it is necessarily problematic.
Some forms of aversive stimuli can, however, generate strong emotional reactions and the resulting arousal and even pain. We might fail to pay attention waiting at a red light and evoke a beeping behavior from impatient drivers behind us. This might elicit emotional behavior that we might experience and label “embarrassment,” which results in our resolving to pay closer attention next time. Depending on our conditioning history, this could be a particularly aversive experience that results in not only the adaptive paying-more-attention behaviors but also rather unpleasant emotional behaviors, which may then generalize to driving in general or even to interacting with other people in general. One might stick a fork into a toaster to dislodge a stuck piece of toast only to be shocked, resulting in injury and a general aversion to toasters or even all electrical appliances. One might contradict a spouse, only to have them lash out at us vocally or even physically, resulting in numerous side effects. Although challenging to operationalize and measure quantitatively, predicting how intensely aversive a stimulus will be, is more or less reliable as well. Failing to reinforce non-criterion responses or withdrawing ongoing reinforcement is aversive, but it is generally not as problematically aversive as harsh punitive stimulation. Noncontingent aversive stimulation is much more problematic than contingently applied aversive stimulation, and readily escapable and eventually avoidable aversive stimulation is much less problematic than inescapable and unavoidable aversive stimulation.
Furthermore, some forms of aversive stimulation are productive (in a literal sense), causing the expansion of adaptive repertoires of behavior, whereas other forms are not productive and only result in the expansion of maladaptive behaviors. Two forms of aversive stimulation might be equally aversive, with the only difference being that one is productive while the other is not. Extinction of a problematic behavior may result in a momentary emotional reaction and resulting arousal, but it may then also result in a decline in the ineffective behavior and perhaps an increase in some more effective (and acceptable) behavior. The relief from stepping forward, and not falling on our faces when walking, results in proper (and effective) walking behaviors. Alternatively, rubbing a dog’s nose in feces as a supposed “punishment” for voiding in the house is unlikely to result in the expansion of adaptive behaviors and is likely to result in secretive voiding and numerous other problematic side effects, both operant and respondent. Growling at a stranger might result in an owner screaming at the dog and yanking their leash, which might result in a decrease in growling but also an establishing operation for escaping strangers in the future, resulting in an increase in alternative behaviors such as lunging and snapping as a prepotent reaction.
It is worth making the distinction between socially mediated and direct aversive stimulation that we could not have reasonably foreseen and/or prevented. The seat-belt buzzer might be aversive but it might also be deemed useful to prompt the securing of one’s seat belt. Some forms of aversive stimulation we deem useful and necessary because we cannot easily identify a less aversive solution. In other instances, a particular socially mediated form of aversive stimulation simply may not be the most reasonable and least aversive solution. Rubbing a dog’s nose in feces is not only unlikely to result in a decrease in voiding in the house but it is also highly likely to result in numerous problematic side effects.
To summarize and conclude this section, in some cases, the aversive stimulation is a direct outcome of the behavior, and in others, the stimulation is mediated socially. For the stimulation that is contacted directly, some forms are useful and minimally aversive/problematic and others particularly aversive and problematic. In this latter case, if we can reasonably foresee and prevent the aversive stimulation in favor of a less aversive solution, this is preferable (i.e., avoidance). In cases where we mediate the stimulation, some forms may be minimally aversive, productive, and necessary, while others are particularly aversive and problematic, unproductive, and unnecessary. For instance, extinguishing a problem behavior is usually necessary, productive, and minimally aversive, particularly when an errorless approach is used and a replacement behavior is installed that serves the same function. Rubbing a dog’s nose in feces, on the other hand, is a different matter. That would be intensely aversive, problematic in terms of side effects, unproductive in terms of solving the problem, and unnecessary in that other, better solutions, exist. This is why the topic of aversive stimulation is not as simple as whether the stimulus is aversive per se or not. Aversive stimulation per se is not the primary problem; the problem is aversive stimulation that is unnecessarily intense and/or unproductive.
In the examples of putting your back foot forward to walk and avoid falling, turning the steering wheel when reaching a curve in the road to avoid crashing the car, or putting on a seatbelt to escape the aversive buzzing sound, “these are not the kinds of things I am opposed to” might have been evoked in the reader and that is exactly the point being made here. Railing against “aversive stimulation” is not exactly the most productive solution, as it misidentifies the problem. The problem is aversive stimulation that is intense or harsh where it does not need to be and/or is unproductive where more productive solutions exist. For example, pinching a dog’s ear to generate a sit that can then be subtractively (i.e., negatively) reinforced with release of the ear is completely unnecessary. It is harshly aversive and highly likely to generate serious and intractable side effects. There are much more productive and less aversive (“best practice”) methods available. Even if there were no less aversive “solutions” to generating a sitting behavior, it is unlikely that such methods could be justified simply on the basis of training a dog to sit on cue. The solution is more of a threat to the dog’s behavioral well-being than the problem of not sitting on cue. As we explore the topic of aversive stimulation that we participate in mediating or allow to happen directly where we could have prevented it, continue to think about the (a) necessity, (b) intensity, and (c) productivity of the stimulation. If such stimulation is not necessary in the sense that a less aversive solution exists or the risks outweigh the benefits, then the stimulation is not justified. If such stimulation is not productive, then it too is not justified. Rarely is intensely aversive stimulation, the kind that is likely to cause problematic side effects, justified—this would represent an extreme scenario that is usually avoidable with added (i.e., positive) reinforcement-emphasized methods.
There are several algorithms, flow charts, and models available that provide guidance to trainers on how and/or when to implement more intensely aversive methods in their training plans.1 These resources commonly recommend minimally aversive added reinforcement-emphasized methods to start, and when the changes in behavior are inadequate, the algorithm justifies an incremental increase in the intensity of aversive stimulation in the contingency management plan, followed again by the solution of greater levels of aversiveness if the technologist fails again. The solution justifies a ratcheting up of levels of aversiveness.
Although it is appropriate to limit increases in aversiveness, and indeed, at some point, a slight increase in aversiveness may be required in some cases, this aversiveness-ratcheting strategy ignores the actual variables resulting in the inadequate training, thus leaving ratcheting as the only solution. In most cases, this aversiveness-ratcheting strategy makes the false assumption that increasing the aversiveness of a contingency management plan is the best solution when an intervention fails to generate
1 This includes models I have previously published (O’Heare, 2013). These aversion-ratcheting models often have the phrase “minimally aversive” or “minimally intrusive” in their title, alluding to aversion-ratcheting as a solution to failure. That is also why I am referring to the strategy provided here as “added reinforcement-emphasized”—to clearly put the focus on added reinforcers. Aversion ratcheting models are the paradigm at present and have been for quite some time. The insidious implication they perpetuate is unstated and it rarely occurs to the authors that these implications exist. Indeed, the authors’ “intent” is to help ensure that fellow behavior technologists use as little aversive stimulation as possible, a laudable goal to be sure and a primary reason why the paradigm persists. I humbly hope that stating these assumptions and implications clearly and proposing a strategy based on a different kind of solution will help contribute to a paradigm shift in this regard.
adequate results. Indeed, the authors of such models may only be aware that they limit aversiveness and be unaware of the implication that the best solution is to ratchet up aversiveness.
Where the actual problematic variables are ignored in such models, if one fails to achieve their objectives with an added reinforcement-emphasized plan, the aversiveness-ratcheting strategy provides the choice between failure on the one hand and increased aversiveness on the other. This leads to the implication that a lack of aversiveness is actually the problem. Of course, the best solution to a lack of aversiveness is not an incremental increase in aversiveness.
The actual problem is some misstep with the assessment, contingency management plan, or its implementation, and the actual solution is to identify and resolve the misstep. The failure of an intervention to generate adequate changes is not necessarily due to a lack of more intensely aversive stimulation, nor is an increase in the aversiveness of the intervention the best solution.
This is not to say that aversive stimulation is never justified. However, if an intervention is not generating adequate changes, the best, most productive solution is to identify exactly why and change that. An intervention may fail because i) the results of a functional assessment have an inaccurate or incomplete contingency analysis, or ii) of an unsuitable procedure or set of procedures, iii) the implementation suffers from problems, iv) the progress is being sabotaged by out-of-session training (i.e., “bootleg reinforcement”) to the contrary, or v) any number of other problems.
Therefore, again, the best, most productive, solution is to identify the actual source of the problem and make the necessary adjustments. For example, if a training project is not going well, one might recognize that the dog is hyperactive and distracted and this is disrupting the training efforts. The most productive solution is not to increase aversiveness but rather to reduce the ambient distraction and generally ensure more exercise for the dog, changing the motivating operations.
Recognizing the problem is a skill, as is finding a suitable solution. Expanding one’s repertoire of such problem-solving behaviors is more productive than increasing aversiveness. As mentioned, this too is not to say that a small increase in the aversiveness of a plan is never justified. One should always try to find a less aversive solution before resorting to a more aversive solution. However, this is quite different from an aversiveness ratcheting approach, which tends to ignore the real problems with the relevant contingencies. In an aversiveness ratcheting approach, the solution to ineffective training is an increase in aversive stimulation. In the approach advocated here, the solution to ineffective training is identifying and resolving the variables causing the actual problem.
Emphasizing added reinforcementbased methods
Success with added reinforcement-emphasized methods requires proficiency in their application and a dedication to find solutions to problems when they arise. Most technologists that resort to aversive methods are simply not adequately proficient in the application of added reinforcement-emphasized methods, including identifying what variables are causing problems and resolving them. Some technologists exhibit a belief that they must “resort to” coercive methods when they fail to achieve quick initial success without them. Failure to achieve training objectives with errorless added reinforcement-emphasized methods should prompt trainers to identify the inefficient and/or ineffective practices or assumptions causing the difficulty. This allows trainers to make the necessary adjustments to the program, in order to resolve them, rather than “resorting to” more aversive methods that may hide, and often compound, the difficulties. Therefore, if one wishes to utilize an errorless added reinforcement-based approach and minimize aversive stimulation, one ought to first acknowledge that the failure of an intervention is likely the result of problems with the assessment, plan, or their implementation. One ought to work to increase one’s general proficiency with regard to finding added reinforcement-emphasized solutions, and when such failures occur, emphasize identifying and fixing the problem, as opposed to ratcheting up the aversiveness. This general strategy is sure to be more productive.
Why implement the added reinforcement-emphasized
contingency management strategy?
The strategy proposed in this paper is presented because of its careful attention to long-term effectiveness, including the effects on the target behavior, as well as the well-being of the subject in general. What reinforcers are available to maintain behaviors that comport with the strategy presented here? After all, it clearly requires a higher response effort and may indeed limit access to certain short-term, impulsive reinforcers. While intensely aversive methods and the eliminative approach in general may impulsively provide a suppression of the problem behavior, thus allowing escape from the aversive condition it establishes for the trainer, these procedures cause a number of problems that may not be immediately apparent, but will ultimately cause more problems than are resolved. Knowledge of added reinforcement-emphasized methods and troubleshooting methods to resolve progress issues, and a dedication to use the least aversive methods possible, helps avoid progress problems and insidious problem side effects. This expanded repertoire of problem solving and emphasis on avoiding intensely aversive methods brings about a more productive state of affairs in the long run, which is why it is the wiser approach and worth the added effort. The point here is that it may not be evident to all, but the most productive course of action when faced with progress problems is to identify the problem variables and find solutions for those problems rather than mask failure with aversive stimulation that may momentarily make it seem as though the problem is resolved, but will end up causing even greater problems.
When professional behavior comports with a strategy that emphasizes increased skill in identifying and modifying inefficient or ineffective contingency management planning and training practices, benefits accrue to the all involved parties (e.g., subject, client, and the individual trainer), as well as the behavior technology field as a whole. The subject benefits from the standard by experiencing a higher degree of comfort and behavioral well-being that comes with being conditioned to react to stimulation in a way that ultimately promotes an expansion of their repertoire of adaptive social behaviors within the family and contacts a greater number of added reinforcers. The client benefits from the standard by avoiding the necessity of dealing with the well-known side effects that commonly occur with the use of highly aversive methods and their objectives will be achieved in an orderly manner. By providing effective, minimally aversive training, the individual technologist benefits from stronger success rates, reduced risk of injury and liability exposure, increased business due to a good reputation, and the respect and trust of clients, colleagues, and allied professionals. The field benefits from the standard with market growth and increased respect from the public and allied professionals. Notice that these are the same reinforcers available for the adoption of all best practices and high-standard guidelines. In the long term, adopting a high standard of ethical behavior, including dedication to implementing this or similar strategies provides greater benefits to society than the failure to adopt such a strategy.
Bringing behavior under the control of practices described in this strategy tends to generate pride-related feelings and thoughts as well. Increased knowledge and skill make for a much more reinforcing endeavor.
Strategy for avoiding versus banning aversive methods and tools
This paper does not present a list of banned, disallowed, or outlawed specific tools or methods. The strategy described here is simply not the place for this kind of prohibition for the following reasons.
First, this strategy is intended to be comprehensive and all-encompassing and to provide guidelines through all possible situations all the way from simply preempting a problem behavior without even interacting with the subject, all the way to the worst-case emergency scenarios that are extremely rare and even to the consideration of euthanasia. If the strategy is going to be comprehensive in this manner, then it cannot simply stop short and ban intensely aversive methods—that would simply leave people unarmed in terms of how to avoid getting to that point and leave them helpless, when and if such a rare and extreme case does present itself. Although worst-case scenarios will be extremely rare, they are theoretically possible, and the strategy accounts for all possible scenarios. Therefore, it would present a discontinuity to then say that this or that specific tool or method is never ever to be utilized no matter what, period. If the model is going to proceed all the way to the worst-case scenarios, then it must not then prohibit outright, the tools and methods utilized in these extremely rare circumstances. Rather than stop the strategy short of the worstcase scenario and providing an incomplete strategy, it provides the guidance required to minimize aversiveness. The entire strategy is built on the idea of helping the user focus on the real problem and avoid increasing aversiveness. The reason that the contingency management plan is failing is not because it fails to include intensely aversive methods. Behavior is driven by reinforcement and added reinforcers can be harnessed via an errorless constructional approach to generate effective changes in behavior. Furthermore, the solution to whatever the impediment is, involves resolving the impediment or finding a solution that allows the problem to become irrelevant. Helping avoid something by training how to avoid it is preferable to legislating the avoidance precisely because it trains in what “to do” rather than what “not to do,” as some like to put it, something we always emphasize with our clients, because it is more productive. The focus is on arming the technologist with a set of guidelines that will prevent the unjustified use of any harsh aversive tool or method, as opposed to providing a specific list of tools/methods that may never be considered. Followed diligently, I dare say no technologist would ever get to the stage where intensely aversive methods are considered. The entire model is heavily focused on finding a more productive solution. Make no mistake about it, the strategy outlined here presents a strong stand for utilizing the least intrusive interventions possible, but it does so in a more productive way than specifically outlawing specific things.
Before I proceed to the strategy itself, some trainers have been outraged that I would even include a Box 6—the box that includes intensely aversive methods. To that criticism I will make two points. First, the entire process here is about how to avoid utilizing intensely aversive methods. I believe this to be more realistic and productive than a blanket ban on Box 6 interventions. Second, exactly what will these technologists do when and if they ever do face reaching a Box 6 case? We are talking about when they have diligently tried everything to avoid it and the problem is extremely dangerous and cannot be managed. What do these trainers do then? Simply walk away? Euthanize? Is that really the most ethical solution?
I am told these technologists simply would never reach Box 6 and not banning Box 6
outright will legitimize people moving forward to Box 6. To the first claim, I say great and indeed a competent technologist proceeding diligently through this model will not likely reach Box 6. However, it remains a possibility, and what then, which brings us to the second point. To that second point, guidance is more important than my taking responsibility for incompetent “professionals” reaching Box 6 without appropriate diligence. Indeed, just about the only way that one would likely reach Box 6 is not to follow the strategy, and of course that is not the fault of the strategy, author of the strategy, or any organization that elects to make the strategy a policy. One can be strongly dedicated to added reinforcement-emphasized methods and allow for the possibility that the use of some aversive methods might potentially become necessary and indeed the least harmful solution left to consider.
The strategy presented here emphasizes three primary systems:
- Objectivity and accountability through proper measurement;
- Emphasis on errorless constructional approach; and
- Reaction to failure of identification and resolution of problematic variable(s).
First, the strategy emphasizes objectivity and accountability through establishing precise quantitative behavior objectives and careful quantitative tracking of the behavior throughout the process.
Second, the strategy emphasizes the use of an errorless constructional rather than trial-and-error eliminative approach as possible. A constructional approach involves increasing the subject’s repertoire of adaptive behaviors. The errorless approach involves arranging the environment in such a way that acceptable replacement behaviors are most likely to occur as opposed to problematic behaviors allowing for a strong history of reinforcement for that new adaptive behavior, which then replaces the problematic behavior. Arranging the ejv9onemnt is this way usually means breaking tasks into smaller more manageable projects and presenting relevant stimuli in a graded manner so as to prevent errors. The eliminative approach involves placing the subject into a trial-and-error environment and the elimination of problem behaviors with punishment and extinction (Goldiamond, 1974/2002; Delprato, 1981).
Third, failure to achieve the objectives prompts careful reevaluation of the behavior objective, the contingency analysis, the choice of procedures, and implementation-related variables.
Failure to identify and resolve the problem may prompt an evaluation of the strategies and procedures utilized and finding a more successful way to put the subject in a position to exhibit the replacement behavior rather than the problem behavior. It may also prompt a change in the motivating operations and other antecedent conditions differentially setting the occasion for problem behaviors versus replacement behaviors. This reaction to failure does not involve the application of particularly harsh aversive stimulation, but rather reevaluation and problem-solving measures.
A proficient animal behavior technologist should be able to plan and implement an errorless added reinforcement-emphasized contingency management plans and completely avoid harsh aversive stimulation. Where a technologist is frequently faced with difficulties in achieving training objectives with added reinforcement-emphasized methods, the best solution is, again, not to resort to more aversive methods but rather to increase their own repertoire of effective planning and implementation of added reinforcement-emphasized training, and more effectively identifying and resolving problems when they face them. The solution is education, not coercion.
The flow chart in Figure 1 depicts this process.
Figure 1. This algorithm provides guidance on how to identify problems in training plans and make adjustments to help achieve success in achieving behavior objectives.
Box 1. Functional assessment and behavior objectives
The first step in the strategy is to carry out a formal functional assessment of the problem behavior and any potential replacement behaviors, and to identify and operationalize specific target behaviors and quantifiable behavior objectives with respect to the problem behavior and replacement behaviors. Without clarity, specificity, and objective accountability, success will be less likely.
The behaviors involved in the contingency management plan are defined operationally and functionally. Operationalizing the target behavior involves describing it in a manner that is directly observable and quantifiable/measurable, not vague or speculative. Defining a target behavior functionally involves defining the behavior by its function—the reinforcer that maintains it. The contingency analysis is not a broad, generalized structurally/topographically oriented diagnostic label, but rather an accurate, reliable set of contingency diagrams describing the specific target behavior and the independent variables influencing it (i.e., the function-altering and evocative stimuli preceding the behavior and the added or subtracted reinforcers follow the behavior). The functional assessment leads scientifically to identification of these variables, and the contingency analysis sums them up concisely.
Once the evocative stimulus and any function-altering stimuli and the consequences (i.e., specific reinforcers) that are maintaining the target behavior are known, the technologist is in a position to develop a strategy, set of procedures, and plan that will manipulate the evocative stimuli, function-altering stimuli, and the consequences so that the behavior will change. Where problem behaviors are involved, the goal is to make that problem behavior irrelevant, ineffective, and inefficient (O’Neill et al., 1997).
Box 2. Construct and implement errorless +R-emphasized contingency management plan
In this phase of the project, the contingency management plan is constructed. The plan includes the objectives for the program, the errorless strategy and specific procedures that will be implemented, and any implementation related details or instructions that may be required. The contingency management plan is not a hodge-podge of anecdotally supported intuitions or “hit or miss” “tricks of the trade,” or the result of trying just another “tool” from a “tool box” of such tricks. The contingency management plan is an evidence-based application of well-established strategies and procedures well supported in the natural science literature. Utilizing a natural science based approach makes it far less likely that one will meet with difficulties and hence a supposed need to formulate a more aversive approach. Once the systematically constructed training plan is implemented, the target behavior will be tracked quantitatively throughout the case to ensure objective accountability.
First, consider whether antecedent control alone could provide a resolution to the problem. Preempt problem behavior by manipulating evocative stimuli, prompting replacement behaviors, and preventing ones that evoke problem behavior. Manipulate motivating operations, perhaps utilizing noncontingent functional reinforcement (i.e., presenting the reinforcer that maintains the behavior on a random time-based schedule, but not after the target behavior) to eliminate the establishing operation for the target behavior. Manipulate other function-altering stimulation, to promote occurrence of replacement behaviors instead of problem behaviors, addressing variables such as medical conditions, nutrition, physical stimulation, stress-inducing environmental stimulation, and so on, such that problem behaviors are less likely to occur.
Where a more invasive contingency management plan is required, utilize all of the above antecedent control measures and complement them with added reinforcement-emphasized postcedent control measures as well. This usually involves a graded differential added reinforcement procedure. Gradually replace the problem behavior with an acceptable alternative behavior by prompting and additively reinforcing it in the appropriate environment and manipulate that environment to ensure success. Utilize an errorless approach by breaking the project into smaller manageable steps, and incrementally and gradually increasing the level of intensity of exposure to problem stimuli such that the subject does not sensitize or exhibit the problem behavior. In addition, it is best to train different components separately, in order to make success more likely. This might involve training replacement behaviors completely outside of the problematic setting to fluency, and then gradually introducing the subject to the problematic setting while manipulating such variables as distance, duration, and distraction. The errorless approach should minimize occurrence of the target behavior, allowing for the installment of the replacement behavior in as close to an errorless manner as possible.
The errorless approach ensures that as few instances of the target (i.e., problem) behavior occur as possible and in some cases, this can mean that these “errors” never occur. In many cases, the replacement behavior can be completely installed without any such mistakes. However, if the target behavior occurs, it must be extinguished, that is, it must not contact reinforcement if it is to be effectively displaced. These “mistakes” should also, of course, prompt evaluation of what went wrong with the plan or its implementation. Extinction is an aversive contingency, but by using an errorless approach, manipulating the motivating operations, and installing an acceptable replacement behavior that accesses the same reinforcer early on, the aversiveness is minimized dramatically.
Furthermore, while extinction is aversive, it is at least productive, and indeed necessary, should the target behavior be evoked. Extinction involves preventing the target behavior from generating the reinforcer maintaining it (i.e., the functional reinforcer). In some cases, other ongoing stimuli may also contribute to reinforce the behavior. In these cases, while the functional reinforcer is withheld, these other ongoing reinforcers can potentially maintain the behavior through a longer, more protracted, extinction curve. For example, continued social contact may function as a reinforcer in some cases, even if it is not the primary functional reinforcer in the contingency. This is an ongoing reinforcer, present throughout the session. Where necessary, it can become important to avoid any potential reinforcement of the problem behavior, not just the functional reinforcer. In these cases, the ongoing reinforcement is subtracted for a brief period of time contingent on occurrence of the problem behavior as well.
To reiterate, while any and all instances of the target behavior must be extinguished, occurrence of the target behavior should be minimized with an errorless approach.
Box 3. Analysis of failure: Identify and resolve the cause of inadequate progress
A well-constructed and well-implemented contingency management plan designed to achieve realistic goals will usually be successful. Minor problems and their causes may become obvious, and simple adjustments necessary, as you proceed, but there are many variables involved in replacing problematic behaviors, some of which occur outside of the presence of the technologist. Problems can occur and it is not always easy to identify and rectify them. If the problem behavior occurs, this means it was evoked. Have you misidentified the antecedent
® behavior contingency involved and hence the behavior was evoked by a stimulus condition that you did not appreciate was evocative, or did you merely allow contact with the known evocative stimulus? Was a function-altering stimulus, perhaps a motivating operation, the problem? Do you need to review and adjust the contingency analysis? Do you need to carry out functional analyses to isolate and test potentially functional relations? Do you need to find a solution for ensuring the subject does not contact the evocative stimulus? Is the failure due to unrealistic expectations? Are you requiring too much of a change in behavior too quickly? Are you jumping to new criteria levels before you have conditioned the previous levels to steadystate? Are you failing to maintain minimal distraction, duration, and distance to start or combining these variables too quickly? Is the problem just a matter of needing more time to ensure that you are moving at the subject’s pace? Make the necessary adjustments to the plan, including controlling the variables that are causing difficulties. Set the subject up for success!
Exhibiting the behavior means that the subject has to experience an extinction trial, which is best avoided. More important than whether the behavior occurred is whether it accessed reinforcement or not. If the behavior occurred but extinction was in place, at least this will contribute to eliminating the problem behavior. If the behavior accessed reinforcement, this is a much more serious problem. It is vital that if the behavior accesses reinforcement directly or the behavior is dangerous, that the behavior be effectively precluded. If the reinforcer is accessed extrinsically, can be controlled, and it is not dangerous, then it must be extinguished every single time. Failing to prevent contact with the functional reinforcer every single time the behavior occurs essentially puts the behavior on a sparse variable ratio schedule, which may make it much more resistant to extinction. Furthermore, extinction generates topographic variability and usually an increase in the intensity of the behavior. In this way, failed extinction can shape a much worse problem than the original behavior. The situation must be carefully evaluated and specific plans made to ensure extinction for every single occurrence of the target behavior. That is, of course, only if the errorless approach breaks down.
Make sure that the client fully comprehends how to manage the environment in such a way that the target behavior is unlikely and where appropriate, that replacement behaviors are likely. Also make sure that the client reacts appropriately to breakdowns in the errorless approach and occurrence of the target behavior. Make sure they are not reinforcing the problem behavior at all! Explain the risks associated with sparsely scheduled reinforcement of problem behaviors and help them find ways to prevent reinforcement of the behavior. If they have been instructed merely to avoid all instances of the subject contacting problem evocative stimuli, ensure that they have been doing so effectively. If the contingency analysis generated through the functional assessment is incomplete or inaccurate, go back and resolve that deficiency, so that the problem stimulation is clearly identified. Question clients about other aspects the subject’s daily routine to ensure that the subject is not training under stress or emotional arousal (i.e., under motivating conditions that make problem behaviors more likely than replacement behaviors). Have the client demonstrate the training they have been implementing and remediate where necessary. Reconsider what they are capable and incapable of working on between consultations. Ensure that they are not engaging in appropriate training procedures during “training sessions” but then inadvertently counter-conditioning that training in “everyday life.” Take whatever actions are necessary to ensure that the client is implementing the proper training at all times.
Finally, evaluate other implementation related practices and variables. This evaluation process is not a cursory “technicality” in which you recognize only obvious mistakes. If everything is being done right, then you should be achieving success. If problem behaviors are being exhibited regularly or the target replacement behavior is not coming along smoothly, then something is wrong. This is your opportunity to identify that problem and resolve it, rather than resort to more aversive methods and tools. Aversive methods will not identify and resolve the problem for you—they may only hide it temporarily. Again, something is, wrong so if you find nothing wrong, then you are missing the problem as opposed to confirming there is no problem.
Consider either the possibility that you may have misidentified the actual functional reinforcer involved or the motivating operations establishing the effectiveness of the reinforcer. It is a common mistake to fail to recognize function-altering stimulation that sets the context in which the behavior occurs, including motivating operations. Is being “hungry” a necessary part of the problem? If so, feed the dog more regularly and eliminate that motivating operation.
There are many moving parts in a contingency management plan and many of them can derail smooth progress. Are there unrecognized sources of distraction? Are there concurrent contingencies operating on the contingency of concern? Are motivating operations being managed effectively for the target and replacement behavior? Is the intensity of exposure to the evocative stimulus being managed carefully enough? Are the reinforcers being delivered contingently and contiguously?
Is the response effort for the replacement behavior lower than that of the target behavior? Consider all of these and other components of the plan and its implementation to find and resolve any deficiencies that may be standing in the way of progress. Contingency management can be complex in the real world, largely because of the dynamic nature of the environment and the variables that influence conditioning. When a well-constructed plan falls short, this is largely where it does so. It can be a challenge to identify the implementation-related problems. If you have achieved some success, analyze to what this success is attributable. What differs in that situation as opposed to when the problem behavior occurs? Often, video recording the sessions can help you analyze the problem and your approach. Consulting a colleague can also be helpful, as well as provide a fresh perspective on the training plan and its implementation.
Box 4. Escalate efforts to identify and resolve the cause of inadequate progress
If your progress continues to fall short of what should be a reasonable expectation for change, despite your efforts to clearly define the problem evocative stimuli and arrange the environment to prevent contact with them, it is time to take more arduous steps in identifying and resolving the problems. Notice that what this stage justifies is not an increase in aversiveness but rather an increase in response effort associated with identifying and resolving the impediments to training.
Start by recommending a full veterinary examination, including blood work that might identify a medical problem (i.e., a disease process or injury) that might contribute to the problems with training. Various medical problems, ranging from simple localized pain due to a minor injury all the way to serious systematic disease processes, can interfere with training. Proceed with the contingency management plan once the veterinarian has identified and resolved the problem or states that they can find no problem. If you need to work around any particular medical issues, consult with the veterinarian on what to avoid when working with the subject.
Refer to authoritative sources such as articles, books, or videos describing the proper application of the procedures in question. Consult a colleague with relevant proficiencies. A fresh perspective, particularly if the colleague can observe you work with the procedure, either in person or on video, can reveal problems that would have continued to stymie the project and they can provide suggestions to get your plan back on track. Another option is to seek supervision for the case. This option has the added benefit of helping you develop your own formal proficiencies. It is also an excellent way to meet your training objectives, promote your professional development, and broaden your skill sets. It might even qualify for continuing education credits.
If these options are unavailable and you are otherwise still unable to identify the problem, you should consider referring the case to a behaviorologist or technologist with specific proficiencies related to the issues involved in the case. The Association of Animal Behavior
Professionals is a useful resource, particularly as certified members are behaviorologically oriented and specifically dedicated to using added reinforcement-emphasized methods. It is not a “moral failing” to lack proficiency in certain skill sets; recognizing and acknowledging a lacking in specific proficiencies is laudable when it is followed up with a referral to a professional with the required skills.
If you have diligently reevaluated the case and researched authoritative sources; if consultation, supervision, or referral are ineffective or not viable options; and the plan is still not sufficiently effective, you should consider finding a different kind of solution. You can consider selecting a different strategy and different set of procedures or a different combination of procedures or you may choose a different replacement behavior. Consider just how important the objective is to the subject and client. Perhaps it is worth simply living with this “problem.”
If a replacement behavior is vital to the subject’s quality of life, it is time to consider escalation. However, the escalation is not in the
aversiveness of the stimulation used in contingency management but an escalation in the effort, time, and resources expended to achieve success. Consider supervising the client’s training behaviors more closely, perhaps being present for all training sessions and increasing the frequency of supervised training. This will help ensure that the client is well coached, more proficient, and not sabotaging the contingency management plan with their lack of experience. You might consider offering a discount, making this option more affordable in general, or providing extra training time on a pro bono basis. You might also consider arranging for a board-and-train service so that a professional can train the subject, and once you or a colleague has trained the subject, you or your colleague can coach the client on how to maintain the training. These options are more arduous for various reasons, but are worth considering if you have legitimately reached Box 4. It is extremely rare for a professional technologist to reach Box 4, let alone have to move to Box 5.
Does failure constitute an unmanageable and unacceptable safety risk?
Is the problem behavior an unmanageable and unacceptable safety risk? If you have reached the stage where you cannot achieve your goals after careful reevaluation of every component of the case, colleagues and authoritative sources have not been able to help sufficiently, you cannot refer the client to a competent professional with specific skill sets that would make success more likely, and you are simply stymied, you need to consider just how important the goal is before proceeding to construct a more invasive contingency management plan.
Any time you consider implementing any aversive methods, be they mild, such as simple extinction for behaviors maintained by added reinforcement, or intense, such as shocking a dog for exhibiting the target problem behavior, you need to weigh the likely benefits against the likely risks. The question at this stage is, “Does the problem behavior impose an unmanageable and unacceptable safety risk?” By “unacceptable safety risk,” we mean, “Is the behavior likely to cause significant harm to anyone at all, including the subject?” The more likely the harm and the greater the degree of harm that is likely, the more likely the question ought to evoke a “yes” answer. If the behavior is not particularly risky in this regard, the technologist and client should continue to attempt to find a solution in Box 4, but if this is not possible, they can make other environmental adjustments to mitigate the effects of the problem behavior and “live with it.” If the unacceptable safety risk is also unmanageable, then the problem is dire. “Unmanageable” refers to the inability to find an acceptable means of preventing the problem behavior itself or the resulting harm. Usually, one can adjust routines, practices, or physical elements of the environment that will prevent or mitigate the behavior or resulting harm.
Problems raised in the literature as examples of supremely important and justifying aversive stimulation are car chasing or digging under fences out of the yard to chase deer. Indeed, these are both high-risk behaviors. However, neither is unmanageable as has been suggested. Keeping the dog on leash when outdoors, or putting patio pavers along the fence perimeter to prevent digging out are reasonable solutions that cause minimal harm. Such problems as these require revisiting Box 4 and continued attempts to find the problem and a solution. Box 5 is not for such problems; it is for dangerous problems that cannot be adequately prevented.
The best solutions are not always conditioning solutions. Sometimes, the least invasive approach is antecedent control measures, what has been referred to as “management.” People often make restrictive assumptions about what can and cannot be manipulated to prevent or mitigate the behavior. It may indeed be less expensive for someone to buy an invisible fence shock collar system than to have a physical fence erected, but this is likely to cause significant and intractable problematic side effects (see Polsky, 2000). It is important to weigh the alternatives. The riskier the behavior, the more invasive may be the restrictions or management of the environment. Some dogs simply may not be allowed off leash in public or it may be necessary to not even walk the dog in close proximity to others. The dog may have to wear a muzzle. Is the solution more or less likely to be more harmful than the problem behavior and is there a less restrictive solution? These are important questions, which illustrate the idea of balancing likely risks and benefits rather than simply invoking simplistic all-or-none solutions. The technologist must consider the impact of management on the subject and the risk involved. Some restrictions or management solutions may be so invasive and create such a negative impact on the subject’s life that the behavior must be considered as unmanageable, but this must be a carefully made decision.
Box 5. Complete review and consider supplements/medications
If you have legitimately reached Box 5, you have failed to diligently implement a successful contingency management plan to resolve an unmanageable problem behavior that poses an unacceptable safety risk and you have failed all other diligent attempts to identify the cause for your difficulties through all of the means discussed in the previous boxes. Your available options are narrowing dramatically. It is time for a complete review from the ground up. This is your last chance to find and resolve the problem or to find an acceptable workaround management solution to mitigate the potential harm that could result from the problem behavior. Once you have completed a full review of the case, implement any adjustments that this review generated.
If your full review and reevaluation has not generated results that will assure everyone’s safety, it is time to consider some solutions that are more invasive but may provide a productive solution.
Consider consulting with a veterinarian about the possibility of using nutritional supplements (e.g., 5-HTP), medications, or even minor surgical interventions that might make the unmanageable and unacceptable safety risk more manageable and/or acceptable, if not ideal. The extent of intrusiveness must be weighed against the necessity of achieving the goal in the case at hand. A more intrusive solution may be justified for cases where the behavior is unmanageably and unacceptably risky, and less intrusive interventions have been exhausted. For instance, perhaps a 5-HTP supplement or a selective serotonin reuptake inhibitor medication can allow you to get a “foot in the door” with your contingency management plan that was just not possible otherwise. These are not sedative options and usually result in few if any problematic side effects. More invasive medications exist of course; the client and their veterinarian should consider the potential risks. Again, these options are not to be considered lightly.
In some cases, rehoming the subject is a safe alternative to proceeding to a highly intrusive contingency management plan. Often the antecedent stimulus is simply not present outside of the current arrangement or otherwise can be avoided in another home. A common example involves dogs who exhibit aggressive behaviors toward children. A safe alternative may be to move the dog to a home where they will have no contact with children. Rehoming can be stressful in itself, so it must be weighed against other alternatives. This is not a decision to be taken lightly, but it should be retained as an option worth discussing in some dire cases. In reality, this option is rarely realistic because of the risks involved and paucity of homes available for companion animals who exhibit serious problem behaviors.
These solutions are indeed less desirable than arranging the contingencies in order to install an acceptable replacement behavior where once there was a problem behavior. But, where there is a significant safety risk involved and you cannot find a less invasive solution and no one else you have consulted can either then the more invasive solution becomes justifiable, as unfortunate as it may be. Just be sure that you have diligently exhausted less intrusive options first!
Box 6. Emergency: Consider more aversive procedures
If you have legitimately reached Box 6, you have failed to diligently implement a successful contingency management plan to resolve an unmanageable problem behavior that poses an unacceptable safety risk and you have failed in all other diligent attempts to identify the cause of the problem through all of the means discussed in the previous boxes, and supplementation/medication or rehoming are not viable. Your available options are strictured even further. It should be extremely rare for a technologist to legitimately reach a Box 6 dilemma. Assuming you have been diligent and still reached Box 6, the problem is extremely dangerous and intractable. This is a highly unusual emergency! The only remaining potential solution is a contingency management plan that involves procedures that are likely to be productive but are also more aversive. This is not to say that these methods are generally more effective or that they will be effective in any given case. However, they remain the only potential solution. For example, in an escape related case, if a differential reinforcement-like procedure (SR+ for replacement behaviors) has failed, perhaps because of the weaker and programmed added reinforcer being used, then a differential subtracted reinforcement procedure, using the more powerful intrinsic functional reinforcer will be more successful.
Intensely aversive procedures should only be constructed by professionals who are proficient in doing so and should be executed and supervised or reviewed by professionals proficient in their application, as well. Proficiency does not mean a cursory familiarity or self-study, under most circumstances, but a true proficiency—one developed through appropriate consultation, formal education, and/or supervision by proficient instructors and supervisors along with some kind of proficiency evaluation process. The thing about proficiency is that one does not always know the full scope of what one does not know; a professional lacking proficiency is sometimes not aware of the extent of their lacking in a particular skill set, which is why formal instruction is important. Again, although “lack of proficiency” may have a negative connotation in common usage of the phrase, professionally speaking, we all have
various levels of proficiency in various knowledge and skill sets. Quite frankly, there is rarely (if ever any) use for proficiencies in implementing intensely aversive procedures where there are proficiencies in implementing added reinforcement-emphasized procedures. We cannot all be maximally proficient in all areas. Recognizing and addressing our lack of proficiency in a particular skill set is best reinforced; it is not a “moral failing.” If the technologist is not adequately proficient to construct and implement a more aversive intervention, they should refer the case to a colleague who is. Nevertheless, whether a referral is possible or not, a professional who lacks these specific proficiencies must not undertake the task. Supervision or peer review can help you evaluate your proficiency level.
Even where the professional is proficient in constructing and implementing a highly intrusive intervention, they should seek either formal supervision or peer review in the case. Supervision involves having a professional who is more proficient in that particular skill set take responsibility for the decisions of the case and approve your actions in implementing it. Typically, you consult with your supervisor between sessions to review the data, your actions, and what you want to do next. Your supervisor helps ensure you provide the best possible service. This may be done via video conferencing, phone, or even email, where feasible, as long as it allows for effective supervision. This also helps you develop your proficiencies for future cases. Peer review (i.e., “consultation”) involves having a competent colleague review your plans and the results on an ongoing basis with you, throughout the process as required. They will provide a “reality check” and a critical eye to help ensure that you are doing the right thing. In this relationship, you remain responsible for the case, although you take the peer review seriously. No highly intrusive intervention should proceed without supervision or peer review/consultation, or, where appropriate, ethics committee review and oversight. This may seem restrictive, but these checks and balances help ensure the subject is receiving the best possible service, which is good for them, the client, the professional, and the profession as a whole.
The criteria for effective punishment of a problem behavior including contingency, contiguity, intensity, sufficient introductory level of intensity, control of reinforcers, and manipulation of reinforcer deprivation (Chance, 2009, pp. 210-217), or subtracted reinforcement of a replacement behavior must be carefully observed. I will not elaborate here on the criteria, as professionals proficient in their application should be fluent with them and it would require far more space than is available to address the topic properly here. Meeting these criteria is not always possible, and mistakes are common. Furthermore, it is important to remember that side effects are an intrinsic component of utilizing intensely aversive stimulation, common even in highly controlled laboratory settings and they cannot be completely mitigated.
If one has been truly diligent and still arrives at Box 6 (highly unlikely), then this is a dangerous emergency. Even though intensely aversive procedures are not expected to be more effective than added reinforcement-emphasized methods, they are some of the few options left. If they happen to be effective in resolving the problem behavior in any particular case, then they have literally saved the subject’s life. Side effects can then be assessed and a separate contingency management plan constructed to help resolve them and improve the subject’s quality of life. Always select and implement the least aversive procedure possible, even once you feel justified in introducing aversive methods. For example, consider a graded differential subtracted reinforcement of a replacement behavior before considering an added punishment-based procedure.
Once the highly intrusive intervention is carefully designed, review or supervision is in place, and all agree the intervention is necessary, considering the behavior and goals in question, it can be implemented. Only professionals proficient in designing and implementing intensely aversive contingency management plans should carry out the program. This is not something you can generally expect a client to perform, except in certain situations (e.g., where they are carrying out only a small and relatively risk-free component of the program and they demonstrate that they can carry it out properly). The behavior must, as always, be tracked quantitatively throughout the process, so that the effects of the intervention on the level and trend of the behavior can be known and success objectively judged. If the plan is designed and implemented well, the strength of the problem behavior should quickly decline to an acceptable level. Once an intensely aversive plan is implemented and it is determined to be initially successful, the technologist should transition to a less invasive and more added reinforcement-emphasized set of controls, in order to fill the suppression void left by some aversive methods.
Plans must then be made to rehabilitate the harm caused by the intensely aversive methods. This must not be neglected. A full evaluation should take place post-intervention to determine the behavioral side effects that have been generated by the invasive procedures and plans should be made to rehabilitate them.
If the goal is not quickly achieved, move to Box 7.
Chance, P. (2009). Learning and behavior (6th ed.). Belmont: Thomson Wadsworth.
Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis (2nd ed.). Upper Saddle River: Merril Prentice Hall.
Delprato, D. J. (1981). The constructional approach to behavioral modification. Journal of Behavior Therapy and Experimental Psychiatry 12(1), 49–55.
Goldiamond, I. (2002). Toward a constructional approach to social problems: ethical and constitutional issues raised by applied behavior analysis. Behavior and Social Issues, 11, 108–197. Retrieved September 12, 2005, from http://www.bfsr.org/BSI_11_2/11_2Gold.pdf
O’Heare, J. (2013). The least intrusive effective behavior intervention (LIEBI) algorithm and levels of intrusiveness table: A proposed best-practices model Version 5.0.
Polsky, R. (2000). Can aggression in dogs be elicited through the use of electronic pet containment systems? Journal of Applied Animal Welfare Science, 3(4), 345–357.
Terrace, H.S. (1963). Discrimination errors with and without “errors.” Journal of the Experimental Analysis of Behavior, 6, 1–27.
 Note that while a graded subtracted reinforcement procedure is aversive and certainly not the least aversive procedure available for escape cases, if carefully planned and executed in graded manner, it is not generally extremely aversive either. It is raised here as an option not because subtracted reinforcement is only justified in Box 6 cases but to highlight it as an incremental up-step in aversiveness from a differential reinforcement-like procedure that makes use of added reinforcement for replacement behaviors.