In her newly released book, Clinical Behavioral Medicine for Cats and Dogs, Dr. Karen Overall mentions Behavior Adjustment Training (BAT) as a rehabilitation tool for fear in dogs. Unfortunately, the protocol that Overall described as BAT was similar to BAT in some aspects, but nothing like BAT in the parts that matter. I wish I had been consulted before publication.
I have contacted Dr. Overall in the hope that she will correct her errors in the next revision. In the meantime, I have written this article to address eight of the myths about BAT in that book. Overall's book addresses BAT for fear, but BAT is also useful for frustration, aggression not based on fear, and early socialization. I focus on BAT for fear in this article and limit the discussion to dogs, although BAT can be used with many species. I also focus on the standard set-up version of BAT, which is done without food, but BAT has many forms. I use the feminine pronoun "she" for the dog instead of the awkward "he or she."
MYTH #1: BAT is just like CAT.
TRUTH: CAT is only one of the many inspirations for BAT.
In Overall’s book, BAT and Constructional Aggression Treatment (CAT) are in the same section and are described as being basically the same protocol. I think it’s important that everyone understands that BAT is not CAT 2.0. BAT is no more similar to CAT than it is to clicker training or Control Unleashed™. There are similarities between BAT and CAT in that they both use an increase in distance from the triggering stimulus (“trigger”) as a reinforcer, but there are also some subtle and vast differences, as you will see below.
Although Overall does note some of the important differences with CAT, it's unfortunate that they are lumped together, because they are not the same in spirit or practice. The tendency for critics to lump CAT with BAT without thinking probably comes from the unfortunate naming similarity. If you take time to look beyond the acronyms, you will see that BAT is a modern, dog-centered training technique worth adding to your training toolbox.
MYTH #2: BAT is too trendy and not really new.
TRUTH: BAT is popular because it works. Empowerment protocols are the new face of humane animal training.
The book refers to BAT as a “trendy” technique, which seems to imply that BAT’s current popularity among trainers and pet owners is a problem. Calling something “trendy” is a way to justify dismissing something new for which we are not ready. In science, we don’t do things just because they have always been done that way. Conscientious trainers and behaviorists look for more effective, less stressful, less invasive techniques in which the animal has more control.
Not doing something because it’s trendy is like not learning how to use a computer because everyone else is doing it. It's a common way to justify things as being too trendy, though: you do still here that argument when old-school trainers refer to clicker training, for example.
BAT should not be dismissed as being the same as what we have done before. People are learning to do BAT because empowerment works and is fundamentally different from how we have been training in the canine world. We are behind science in this one, as Applied Behavior Analysis has long since focused on operant protocols to change the emotional response of a learner. With dog training, we have grabbed onto associating the trigger with food and stuck with that, because that was our first major push into rehabilitation through positive means.
BAT is a systematic desensitization technique, but it is inaccurate to write that it is "nothing more than routine desensitization and counter-conditioning (CC/DS)." For one thing, it is probably not classical counter-conditioning because we are not pairing two stimuli. There is classical conditioning, of course, because that is always happening. It is in the form of desensitization.
BAT is a desensitization protocol based on empowerment, where the dog has almost complete control over the process. The control exercised by the handler is only in the form of making changes to reduce stress and ensure safety. BAT adds the feature of ‘controllability’ to the situation. Controllability directly factors in to expression and experience of anxiety—this isn't inherent in standard CC/DS protocols in dog training.
An internal locus of control is a primary reinforcer and most training only gives the dog a token amount of control of their real environment. Through BAT, the handler learns what the dog is "saying" and honors that. While negative reinforcement (R-) probably plays a role, the dog’s control over the process is what helps the desensitization 'stick' (i.e. limits spontaneous recovery). This use of R- is not unethical and I will address that more below.
MYTH #3: People criticizing BAT are familiar with the protocol.
TRUTH: Most criticisms of BAT involve describing BAT incorrectly and then criticizing that incorrect version of BAT.
It’s best to see for yourself. Here is another video of BAT with clips from multiple dogs. I think watching it will help you better understand the rest of this article.
MYTH #4: BAT is just CC/DS as we have always done it.
TRUTH: BAT is a systematic desensitization protocol but it is also operant conditioning with real-life reinforcers. It is an improvement beyond standard ways to do CC/DS for social issues, because it has more empowerment for the dog and fewer distractions.
I developed BAT because I wanted something that would create a more profound change in my dog’s behavior and quality of life than the standard open bar / closed bar version of CC/DS (dog sees trigger --> gets treats) and other operant treat-based versions. I feel that while pairing the trigger with good things is far, far preferable to using corrections and force, it is still actually invasive in the sense of removing control and not allowing the dog to really process the situation.
Food can certainly be helpful to relax your dog, and I use it whenever it's necessary (small spaces, surprises on walks, etc.). Unfortunately, food, handler interaction, or other high-valued items used in most CC/DS protocols also draw too much of the dog’s attention, limiting how much information the dog takes in. Real-world triggers are usually intramodal compound stimuli, i.e., they are made up of stimuli from different sensory modalities. Pairing the category of "all children" or "all dogs" with food in real life is much more challenging than simply associating the sound of Pavlov’s bell with meat powder in a laboratory.
Even with a positive association with some salient aspects of the trigger, the dog may still react to other aspects, because of overshadowing and other factors. In experiments with pigeons responding to intramodal compound stimuli, the reinforcer used in the experiment determined which element of the stimulus received more attention from the pigeon (see “intramodal studies” in this article). In particular, the use of food or other high-value items may determine which elements of the trigger the dog notices and remembers. An incomplete understanding of the trigger may lead to a plateau in training where the dog is only ‘fine’ as long as she is focused on the handler.
In contrast, people who have turned to BAT often report that their dog is now “a different dog” who interacts with dogs or people like they have always dreamed their dog would do. I recommend food-based CC/DS protocols for sound sensitivity and other applications with simple triggers and as a management strategy for other situations. Even then, whenever you can do an operant protocol that brings classical conditioning along for the ride, that's the better choice. For most social situations, physical locations, and compound triggers, I use BAT, because it allows the dog to fully perceive the trigger with fewer distractions.
I have drawn from many resources to create BAT as a humane way to give the animal even more control and awareness than CC/DS. Individual cases are complicated and you can combine BAT with standard food-based CC/DS and clicker training, as needed. Whatever techniques you use, don't forget about empowerment: giving the animal more control over the process (as in BAT) leads to more complete rehabilitation. Empowerment goes beyond simply using operant techniques. Whenever possible, you should take a functional approach that involves choosing a reinforcer that is appropriate in that situation.
For body handling issues (grooming, vet procedures, etc.), dogs should be taught in ways that allow them to communicate when to start and stop the procedures, as is done in many zoo settings and with Chirag Patel’s work for companion animal husbandry. An example would be teaching a dog to rest her chin on a target during brushing. She gets treats for being brushed, but can remove her chin and stop the procedure at any time. Allowing her to manage her own stress level is less invasive than simply pairing brushing with treats. I call the behavior the dog does to signal that CC can continue a "More Please Signal."
MYTH #5: BAT works the dogs over threshold.
TRUTH: A BAT trainer only allows the dog to walk to a point where they are curious, but not over threshold.
Overall pointed out inexperienced handlers can put dogs over threshold with BAT, but that is a risk with any protocol. With BAT, watching the dog for signs of stress is our main focus. Traditional CC/DS can also be done incorrectly, too. In that case, the animal is often over the fear threshold but the person does not notice because they are using 'the dog is eating' as the criteria for being under threshold and there is also a competing reinforcer that influences behavior. BAT clients are able to keep their dogs under threshold during training by learning to watch their dogs’ behavior. Two-way communication is an invaluable skill for anyone with a dog.
The “upsetting stimulus” that Overall mentions is the same exact stimulus used in traditional CC/DS. We usually begin at a much greater distance than with those protocols, because there is no food to distract the dog in a basic BAT set-up. If it is not possible to move far enough away that the dog is below threshold, then we will use Stages 2 or 1 of BAT, which are versions of BAT that use food to help the dog handle the proximity of the trigger because they are in ‘working mode.’
We use these stages only when necessary, because of the distraction factor that I mentioned above. When she is focusing on getting treats from the handler, the dog is no longer able/motivated to fully take in information about the trigger. The dog is much less likely to exchange social cues with the trigger while she is in working mode. Of course, paying less attention to the trigger is helpful for urgent situations and when you first start working up close, but we switch back to ‘regular BAT’ whenever possible for maximum learning.
At any rate, the point is that we are using the same stimulus and the major difference is that BAT gives the dog more control over leaving and approaching while traditional CC/DS provides something appetitive. If it’s the same stimulus, at a further distance, but we are allowing the dog to sniff and look at the trigger instead of feeding her a treat, how could that possibly be less humane?
MYTH #6: BAT just teaches animals to avoid or tolerate situations.
TRUTH: BAT dogs learn to enjoy people, dogs, etc. and know how to reduce stress when needed.
The goal with BAT is not to "teach the dog that he can tolerate avoiding other dogs," as Overall writes, but rather to have the dog able to fully assess a situation and offer pro-social behavior and interaction when she desires and appropriate cut-off behavior in any situation in which she is uncomfortable. As the dog’s ability to control the level of social interaction increases, it seems that she feels more and more comfortable interacting. This is true of humans who learn active coping strategies.
Based on behavior, this appears to be true of dogs and other animals using BAT, as well. So instead of a dog who can pass by a trigger because she has associated it with amazing treats (or worse, fears a correction from a collar), we have a dog who acts like a real dog and is able to truly check things out without the need for micromanagement from the handler.
Overall writes, "If dogs learn to have the expectation that if they just stand there the other dog will leave, we have provided them with some very flawed real-world applications." This may be true of CAT. There is nothing about this statement that accurately reflects BAT, which emphasizes natural behavior and movement of the student dog. The lesson to the dog is “you have the power to de-escalate this situation peacefully.” BAT also includes very careful leash handling skills and long lines/harnesses so that the dogs have the maximum freedom of movement.
MYTH #7: BAT forces dogs to be in stressful situations. The R- in BAT is just like choke chains.
TRUTH: BAT teaches the dog that she can always leave, which allows her to learn more about “real life” and build confidence. Teaching a dog to self-soothe in a low-stress way is empowering and humane.
Overall worries about flooding, which is placing the dog in a situation that the animal would not choose and then prohibiting escape. Flooding can happen with CAT, but not BAT. It is very important to me to build up the concept of choice, not to destroy it. It is really unfortunate that BAT and CAT have been lumped together in Overall’s book and in some people’s minds.
With BAT, the dog is always allowed to leave. We never force a dog to go forward nor do we keep the dog in a place where she does not want to be using force or treats. We walk with the dog only to the point of curiosity and slow her down to a stop. We then wait for the dog to gather information and then disengage from the trigger. We verbally ask "Done?" (I originally recommended using “Yes” as the marker signal) and begin to move away. If the dog does not come with us, we wait for the dog to gather more information and fully disengage, then ask again if she is “done.” If moving away is too challenging, then we take not of that and stop her farther from the trigger the next time.
Overall writes, "even if flooding is not involved, both BAT and CAT are forms of negative reinforcement. Simply, the reward is the release from having to be in the presence of the upsetting stimulus. This is the same principle on which the use of choke collars is based, and these techniques carry the same risks of recidivism or worsening of behaviors that choke collars carry."
There are a lot of points that I'd like to discuss from this paragraph. First of all, flooding is not possible in BAT, because in BAT there is movement of the student dog and the dog is always allowed to leave at any point, no matter the behavior. Unlike in CAT, there is no risk of using extinction; extinction is not used in BAT. If the dog goes over threshold during BAT, we simply encourage her to leave and calm down.
A major concern that positive trainers have is that BAT uses R-, which is not supposed to be done until we've exhausted all other options. One thing to remember is that the primary way that BAT changes the dog's emotional response to the trigger appears to be from a desensitization process that the dog controls. The operant piece is that the association of the trigger with the ability to handle the situation and yes, that's R-. (That's a hypothesis--we don't actually know that reinforcement is happening at all. The great behavior change could just be desensitization or even R+ for looking at the trigger, but it is likely involve at least some R-). There is also R+ in BAT in the form of interaction with the handler and later with the trigger. First the dog learns that she can always move away whenever she wants to, because cut-off signals are reinforced with moving away. Then because of reduced arousal (desensitization) the dog has a chance to experience something pleasant, and social behavior from the student dog is reinforced by social behavior from the former trigger.
Nobody's worried about the R+ aspects of BAT, so let’s look at the R-. In BAT, the R- does not involve the application of an aversive TO the animal—the aversive (the other dog) is self applied as the dog walks toward it. This is different from CAT, traditional pressure-release training with horses, and the use of choke chains.
Furthermore, the aversive used with BAT is the trigger, not an artificial device like a choke chain. It is the same aversive used in all rehabilitation protocols, and usually we start farther away than you would with treats. With BAT, we are not applying pressure to get some unrelated, trainer-desired behavior. We take an environmental pressure that already exists and teach the dog to eliminate her own stress using natural behaviors chosen by her.
We allow her to approach the stimulus that triggers her fear/aggression and retreat from it whenever she wants. The human's role in BAT is primarily to gently stop the dog at the point of curiosity, before she goes too close—not to pull her toward the trigger. We neither train any specific behavior that we desire, nor do we create pressure to get it. We are not teaching the dogs merely to escape, but to thrive because they now have coping skills. They have to live with dogs/men/kids in their world and BAT teaches them to do that in a very natural way, driven by the dog's own curiosity. BAT is dog-centric.
There is no positive punishment aspect. If there were, the approach behavior would decrease, but it doesn't. A BAT dog approaches more and more and begins to offer exploratory and pro-social behavior, indicating that the trigger has become less aversive. To my knowledge, there is no research on this kind of R- and claiming that it has the same effects as choke chains is not justified.
MYTH #8: This is stressful for the decoy dog.
TRUTH: BAT should be low-stress for all animals involved.
Another issue Overall writes about is that the decoy/helper dog would feel stressed. I agree that this is important. I never support the use of a stressed dog, regardless of the training role. With BAT, the wellbeing of the decoy is an important component of the protocol. It seems very irresponsible to me to allow the decoy to have to stand there while the student dog barks, lunges, etc., as in CAT or correction-based training.
As with any systematic desensitization protocol, the handler of the helper dog should monitor stress. BAT takes care to minimize stress in all of the dogs and even the humans within the situation. A relaxed, non-threatened decoy is key as he aids the student dog in offering up her own good choices in pro-social behavior. Because of the fluidity of BAT, the decoy dog’s movement does not impact the core principles of BAT. The decoy dog can be doing BAT herself, getting treats, or doing something else to make the situation pleasant.
CONCLUSION When Karen Overall takes the time to learn about BAT, I believe she will realize that she has misinformed her readers, and will change the assessment of BAT in her (otherwise fabulous) book.
To those of you who haven’t yet learned about BAT because you are concerned about ethics and quadrants: Please take the time to understand BAT and the power that it gives to dogs and other animals. You will find it that it meshes well with other positive training techniques that you already use.
When you actually do BAT yourself you will see where the ‘trendiness’ comes from: empowerment training is humane, makes a lot of sense, and is very effective. Thank you for reading this through!
Please comment below and feel free to share on Facebook. 🙂