What Makes BAT Work?

I’ve done some soul-searching about what makes Behavior Adjustment Training (BAT) work and my hypothesis from a behavior analytical perspective is that BAT involves desensitization, naturally occurring negative reinforcement (R-), and both naturally occurring and trainer-delivered positive reinforcement (R+). I now believe that the change in emotional response in BAT 2.0 is largely due to the controllability of the situation and desensitization/respondent extinction, not behavior built up through reinforcement (although those are always intertwined).

But why am I not extremely concerned that it may have some flavor of R- when I am a force-free trainer? Actually, it did concern me, but it doesn’t any more, because I have thought it through.

I take humane training very seriously and adhere to the Progressive Reinforcement Manifesto written by trainer Emily Larlham, where it is of the utmost importance that life of the dogs we care for is arranged to minimize stress to maximize learning in a way that respects the dog.

I also love Susan Friedman’s Humane Hierarchy. [click here for an interview with Susan]. One of the main points of her guidelines is that trainers need to look closely at any use of aversive stimuli and watch out for things like negative reinforcement.

As Dr. Friedman told me herself, the naturally occurring reinforcement found in BAT is not really part of the humane hierarchy. It’s not an aversive delivered or orchestrated by the trainer, it’s totally different from an aversive like a prong collar or e-collar. It’s just real life to allow dogs to move around. Giving dogs a way to control their own distance to stimuli, to seek their own relief, is a GOOD thing. 

In short, that means we can all stop stressing about BAT on social media and get back to working with dogs.

Any negative reinforcement found in BAT is simply a consequence of giving the dog more control. It’s similar to the control dogs get in any positive technique, but better!

The only way for a protocol in which we allow dogs to notice their triggers to NOT have any R- is to take away their control. That’s not humane.

In BAT, moving away should really be ‘moving on,’  not necessarily avoidance. Moving on after engaging with the trigger can just be satiation, they have had enough contact and are ready for something else. If you are getting complete avoidance of the trigger, you’re doing something wrong; that’s not BAT. We should be working at the level of curiosity, so the movement away is only because the curiosity has been satisfied. Controllability is a good thing and having it around their triggers is especially good. If you do BAT right, most of the movement done after looking at the trigger should be just due to the dog being ‘done’ looking. They are moving on, not escaping.

I think it is important to put ego aside and really assess what people say to see if it holds any truth, and then use that to improve the way I teach dogs. Here’s what I know at this point in time. 1. Allowing the dog to behave in a way that helps them feel more comfortable (naturally occurring R-) is not the same as applying an aversive stimulus and removing it when the dog does what you want (R-) and 2. we do not yet have any techniques that avoid this aversive (the other dog), so allowing the dog to move away whenever he wants is a humane choice.

There are no unnecessary aversive stimuli in BAT, in stark contrast to the kind of R- techniques that use choke chains or electronic collars. The aversive aspect is just the decoy, at a level of engagement that the dog controls. This is less aversive than many techniques, if you really look at the lack of controllability by the dog, and no worse than the best ones out there. The dog has a huge amount of control over the BAT process. For a particular dog, you may have other tools you try before BAT, but a good trainer can always use more force-free techniques.

For most of our BAT set-ups, we don’t use much food (we do use food on walks). In the set-ups, we use the same trigger as techniques that use food or toys, but we are farther away at first. Without a competing reinforcer (or worse, a threat of punishment), the dog’s behavior expresses his internal emotional response to the trigger more clearly, so we can work at the right distance and use a natural reinforcer appropriate to the situation.

We try to consistently work at a point where the dog looks at the trigger, thinks “no big deal,” and turns away out of boredom. We do not put the dog into that position; it is the dog’s choice to move around. Our job is to keep the dog from getting too close. If the dog chooses to walk away then, that behavior is reinforced by whatever comes next; that’s the Premack Principle and probably R+. If the dog is concerned, which he is bound to occasionally be, then he looks and is either relieved by what he learns (R- for looking) and/or the trigger is still somewhat aversive and he is relieved when he walks away (R- for disengaging). Or maybe he doesn’t care one way or the other and there is no reinforcement.

A common misconception of BAT is that the dog is getting a ton of reinforcement for leaving, that the protocol is about escape. The key to BAT is that the dog is learning about the trigger itself, in a safe environment, where the dog can always move away.

With the most common techniques using counter-conditioning, the trainer works close enough to the trigger that the dog will look at it occasionally, which means that even with the draw of tasty food, the dog still needs to occasionally check out the trigger.

To have that kind of pull during CC, even with the possibility of fabulous treats, the trigger is probably aversive on some level even when working below threshold.* If you walk the dog away from the trigger at that point, there’s probably relief involved, and if you do it at the right time, it’s probably still negatively reinforcing. Staying and feeding keeps you out of the R- quadrant, but that’s because you didn’t reinforce any behavior–it’s still an aversive, but the dog has no control over it.

I find that it can be harder to know whether the dog wants to move away from the trigger when there is a competing motivation to stay (trainer with food).

Moving away may be what your dog actually needs in that moment, but how do you know? Did he turn away from the trigger or get between you and the trigger because he wanted a treat or because he wants space? It can be extremely hard to know when your dog is truly under threshold when he is in working mode, because his behavior is being influenced by the presence of the food or toys. When in doubt, assume he wants more space.

I did not discuss this to put down open-bar counterconditioning, but to explain that BAT is at least as humane as those techniques and hopefully put such arguments to rest. BAT and CC/DS are lightyears ahead of techniques like lifting the dog off of the ground, using electronic collars, or popping with the leash. The important thing is to give the dog control over the process and to avoid any aversives that we can. We are all here to improve the lives of dogs and should all be carefully monitoring their stress levels.

If you choose to do BAT set-ups, awesome. If not, that’s fine, but I still urge you to be really, really aware of your dog’s needs. They have a conflicting motivation if you are training with food (or corrections), so you may not be able to read your dog as well.

At any rate, do not be afraid to move your dog away when he needs more space, even though that may be (gasp) negative reinforcement for a cut-off signal.

* Not sure it’s aversive at the distance you’re working at with food? When you are at the right distance for your technique, get rid of the treats and just relax so that your dog is not in working mode. Let your dog check out the trigger without distraction from you. Wait and see what behavior the dog offers. You may be surprised.