As I mentioned last week, my trainer Crystal suggested that Frankie and I try Behavior Adjustment Training (BAT) — not only because it’s a good technique but because it’s a good fit with Frankie’s diabetes. That is, although the technique can include food rewards, it doesn’t rely on them.
I wanted to learn more about what exactly Frankie and I were getting into. Because Grisha Stewart, BAT’s creator, has a lot on her plate at the moment, her fellow trainer and BAT expert Irith Bloom agreed to fill in and answer my many questions. This is the first part of my interview with Irith; there will be at least two more installments.
1. Could you briefly describe the theory behind BAT?
To quote Grisha Stewart, “BAT uses functional and bonus rewards, combined with clicker training and systematic desensitization, to help dogs make better choices in an error-free way.”
In plainer English: Animals, like people, try to control their environment using a variety of behaviors. They generally decide which behaviors to use based on what has worked before. For example, some animals have learned that aggressive behavior makes scary things move away from them, so when faced with a scary thing, they act aggressive.
The goal of BAT is to teach the animal a set of more acceptable tools for dealing with situations they find stressful, and to which they currently respond in a way we humans don’t particularly like.
What kind of techniques does BAT use?
BAT includes a few different techniques but the most important one involves artificially created situations (setups) where the trainer can control how stressful the situation is for the animal. In these setups, we train just at the edge of the animal’s comfort zone, and wait for the animal to offer an acceptable behavior, instead of the one we don’t like. When the animal offers an acceptable behavior, we “mark” that behavior with a clicker or other marker — we tell the animal “that’s it!” in one way or another — and then give the animal whatever it was the animal wanted in the first place.
Here’s an example of how a setup might work:
Fido is scared of men wearing hats. Whenever Fido sees a man wearing a hat, he starts barking and lunging. When he does this, the man with the hat usually backs away, and Fido’s owner also tends to pull him away. End result? Fido winds up farther away from the man with the hat.
In working with Fido, I would bring in a friendly male volunteer who is wearing a hat (the decoy). I would figure out how far apart Fido and the man must be for Fido to notice the man, without being so stressed that he starts barking and lunging. I would then begin with Fido and the man at that distance and wait for Fido to offer an acceptable behavior ([Note: I'll be defining and giving examples of acceptable behaviors next week]). As soon as I saw that acceptable behavior, I would mark it and have Fido’s owner move Fido away from the man with the hat.
I’ve shown Fido that acceptable behavior gets him what he wants (or at least, what I think Fido wants). Assuming Fido wants distance from the man with the hat, Fido will be likely to offer the same behavior again to see if it results in increased distance a second time. (Note that if Fido doesn’t start to offer the acceptable behavior more promptly after a few repeats, I would begin to wonder if I had misdiagnosed what Fido wants, and review the situation to figure out what Fido might want instead.)
We would repeat this exercise until we got a variety of acceptable responses from Fido, and then gradually decrease the distance to the man with the hat. Variations might include having the man walk away in response to acceptable behavior (instead of Fido walking away), or having the man speak to or reach towards Fido rather than standing still.
Note that we would only practice one variation at a time, and each at a very safe distance before the distance was decreased with that new variation. Once Fido was relaxed even up close with the decoy doing strange things, the next step would be to get another friendly male volunteer, put him in a hat, and start all over again with that new decoy.
After going through BAT training with two or three different decoys, most dogs begin to offer acceptable behaviors fairly quickly even when encountering a completely new decoy for the first time. For some animals, one decoy is enough to establish a new set of behaviors (though this is rare). For other animals, more different decoys are required. The number of decoys required depends on a variety of factors, including the animal’s inherent personality, how deep-seated the anxiety is, and how carefully the training is managed.
Being careful not to add too much difficulty at any one time can prevent the wrong kind of responses. When we do everything in a slow, careful way, we can achieve “error-free” learning where the animal never again has an over-the-top response to the trigger.
Irith Bloom is the owner of The Sophisticated Dog, a pet training business offering services to clients on the Westside of Los Angeles. She specializes in clicker training and other pet-friendly methods of animal training and behavior modification. Irith is a Karen Pryor Academy Certified Training Partner) and a member of the Association of Pet Dog Trainers (APDT). You can see more of her writing on her Sophisticated Dog website, on Karen Pryor Clickertraining’s website and on Examiner.com.