OB170: Bridging the Gap - Reducing Reinforcers, Proofing and Generalization

And...again we change direction! This class is made up of three parts - Reducing reinforcers, generalization and proofing. We've introduced the concepts of reducing reinforcers and generalization - those two are tied together through the process of engagement training. Now we'll consider proofing - the final piece.

Proofing is the process of making your dog strong in the face of distractions or alternative interests. For example, if you enter the obedience ring and the steward is eating a muffin, then your dog has to work in spite of an appealing option nearby – the choice becomes between working with you or thinking about that muffin.

Notice that I said, "thinking" about that muffin, not necessarily looking at or trying to eat that muffin. If I have a leash and a collar, I can easily teach the dog not to "look" at the muffin by applying a collar correction. But a correction doesn't change what the dog is choosing to think about and dogs that are thinking about muffins aren't giving me very high-quality work.

You can never change what a dog is "thinking" about unless the dog chooses to make the effort to focus on what you want.

It's like putting a kid in their room a punishment for some misdeed, and then telling them to "think about what they did wrong." I heard that phrase often as a child, and most of the time in my room I spent being resentful and thinking about how I would not get caught the next time. I certainly didn't spend energy on being grateful that my parents were raising me to be a good person. My parents could change my behavior, but not what was in my head.

The problem with a correction based strategy is that a high percentage of dogs never stop thinking about the muffin. They either develop fearful behavior around things that they want (fear of correction), opportunistic behavior (become more clever about getting what they want, especially off leash), or they proceed to work, but the quality goes down because their attention is split. And of course, there are those lovely dogs that always respond well, no matter how we train them. Since the last group is in the minority and training with corrections is not within the philosophy of the school, we'll ignore that one and focus on our other options.

Excellent proofing done within an R+ framework addresses the issue of proofing differently. Instead of preventing the dog from looking at the muffin, the dog is made aware of his choices and possible consequences. He is allowed to try out all of the ones he may wish to exercise. What are his choices? I can think of two:

Dog's behavior (Option #1): Attempt to leave the handler and lunge for the muffin.

R+ Solution: The persona holding the muffin prevents the dog from getting the muffin with the minimal amount of interference. If dog persists to the point of being obnoxious, he will be removed from the situation and crated. Regardless, the choice to lunge never leads to a muffin. This lesson would be repeated until the dog no longer goes for muffins - good choices (looking at handler) should be rewarded generously.

Why it works: Lunging does not work. The dog neither gets the muffin nor does the dog get anything from the trainer. Checking in with the trainer with tasty muffins nearby does work - either the dog is rewarded directly for that choice (ideally an identical muffin from the trainer), or allowed to work (which can lead to reinforcement, possibly a muffin)

Dog's behavior (Option #2): Stare at the muffin.

Solution: Fine. Stare. After some amount of time, the dog is either removed from the situation or...he gets bored with that activity. If his training has been choice based, then he should think to check and see what his handler is up to at some point. If he checks in with mom, he will either be rewarded generously or allowed to work (which leads to possible reinforcement). I can be very patient with this type of dog - no worries if the dog chooses to stare for several minutes. When you do this, make a point of standing behind your dog as opposed to next to him; looking at you should be impossible if he is also trying to look at the muffin. You are forcing a choice.

Why it works: The choice to stare at the muffin never leads to a muffin. The choice to check in mom and give up on the muffin leads to reinforcement or a chance to earn reinforcement (for a more advanced dog).

Imagine that your dog NEVER received a morsel of food from your table. He might hang out with you for awhile, but eventually, he'd probably look away. If at that exact moment, he received a morsel from his food dish, it would take less than a week before your dog would hang out at his food dish when you ate dinner. This would be true even if only one or two pieces of food showed up in that dish throughout your entire meal. SOMETHING from the food dish has been conditioned when the family sits down to eat. The trick here is that the dog NEVER gets anything from the table.

In the above example, the muffin with the steward is the family at the dining table, and the hander with their tasty treats and work represent the food dish of the dog.

How to do it: If you are working with another person, then you can have a "muffin holder" and your dog can be off leash. If you are alone then the muffin needs to sit on a chair or table with your dog on leash, or somehow be inaccessible if the dog makes a poor choice (possibly locked in a crate or an ex-pen)

Allow the dog to stare at the muffin. Say and do nothing. Stand behind your dog. Wait. If the dog turns to look towards you, back up away from the muffin, calling and praising your dog. If he comes all the way in, give the equivalent treat from your body. It's also ok to call the dog away in this manner and then go get the muffin yourself (without your dog) and bring it back to your dog. I rarely send the dog to the muffin, because of the problem with reward for position - the dog often begins to stay half way between the trainer and the muffin. No good!

In this video, you can see that Lyra is made aware of the tasty meat in the bag. I then place it on the bench in front of her and allow her to approach (I haven't asked for anything so there is no reason why she should not approach). I allow her to sniff it - when I realize she's not going to "break into the bag" I loosen the leash. When she returns to me and sets up in heel position (her default), I go get the pork pieces out of the bag and give them to her. On the second attempt, I do the same thing but I already have the cookie in my hand, AND I ask for heeling. Because I have the pork in my hand, I am faster to reward, but if you do this, be aware that it can easily become a lure. Make sure your dog can do both:

In this video, I repeat the exercise, but this time I have a person holding the bag of meat so that the dog can perform off leash with no risk of self-reinforcing. In this case, Raika is also asked to work a bit after she shows me that she can perform the basic exercise with relative ease:

In this final video, I"m working with young Dog Brito. He has to make a choice between working for me (retrieving or putting his feet on a target) or staring at/sniffing the food bowl. Because the food is in a tube, he cannot get to it. Note that I make no effort to stop him from checking. With time, he won't bother if he does not win.

Considerations:

This method will not work particularly well under a few circumstances:

1. Dog has learned that after the handler gives a cookie, he can run off and do as he pleases (which may include trying to get to the proof). If this is your situation, then you'll have to rethink your training. In your case, never reward the dog with the proof - only the equivalent thing from your body. If you do choose to use the proof as the reward, get it yourself and bring it to the dog - do not let the dog accompany you or take it himself (I show this method in Lyra's video above). If your dog continues to leave you after receiving a reward from you, then you'll have to ask the muffin holder (or distraction person) to gently take your dog by the collar and crate him if he goes to them. If you are alone or if your dog cannot safely be handled by a stranger, then let him drag a leash, and you'll get to put him in his crate if he leaves you and runs for the muffin. One minute of crate time is long enough. Try again with the game. Dogs that show this behavior have developed a behavior chain - do something from the trainer and then do whatever they please. You are going to change that.

2. Dogs that are in avoidance will not succeed at this game - if the dog is looking at other stuff (not the muffin) then you probably have a stress problem or a dog who does not care about the proof. For this to work, the dog has to want the thing that you have planted for this exercise. Try to find something else. If you know that your dog wants what is out there, but steadfastly refuses to look at the trainer, then reward the dog for looking away from the proof but without waiting for the dog to check in with the owner. In this case, when the dog looks away, the handler should "mark" the dog's choice to look away from the proof and then "reward for position" by giving the cookie/reward close to the handler. Work on this until the dog anticipates that the treat will come from the front of the handler, and therefore turns from the proof and looks at the handler for the reward - then proceed as above, expecting the dog to turn to the handler rather than simply away from the proof.

3. With a dog that has had traditional proofing (corrected for looking at the proof), the dog knows not to look towards the proof, but they can show stress in the presence of distractions because they expect to be corrected. To remedy this, simply reward the dog as if they had made the choice not to look at the proof. If the dog ever does make a mistake, allow the above chain to take place - ignore the dog's choice to look at the distraction and reward when the dog makes a good choice to look to you. As long as you control access to the distraction, the dog cannot self-reward, and over time the dog will relax in the presence of distractions - seeing them as positive challenges to be overcome rather than possibilities for correction.

Good trial preparation must include thoughtful proofing, until eventually it becomes a way of thinking and training – for both you and the dog. The earlier you start in the dog's career, the easier it becomes for both of you. Do not avoid a challenge if you believe your dog can succeed - meet challenges head-on. To be successful, your dog must believe that you control ALL distractions - if you set them up regularly, cooperation with your will become a habit, well past the point when you do NOT have control - like the steward eating a muffin at the dog show.

Proofing should both strengthen your dog's understanding of and willingness to perform the exercises, even when attractive alternatives exist.

The example given above uses an "attractor" to teach the dog to focus on the trainer even in the face of distraction. There are many other additional types of distractions that can be handled with proofing of various types: novel environments, dogs, people, interesting objects - always neutral or attractive to the dog. In every case, the dog is prevented from getting any satisfaction from the proof. When the dog recognizes this and turns back to the handler, then the party begins.

If you think about it for a moment, the generalization exercise presented earlier in class (dog is allowed to stare at the environment but not interact) is a form of environmental proofing. The dog believes that you can prevent access and develops a habit of working in new places - that habit holds up well past the time when the leash comes off and you really don't have control.

Again, you can see the relationship between generalization, proofing and reducing reinforcers - they are interconnected.
Proofing should go on for the life of the dog. It is a regular part of life and a part of training. I have tremendous patience with proofing - I simply wait the dog out and make sure they cannot get to the thing they want. From there, it is up to the dog to choose.

A word about fear: Fear is never dealt with through proofing. Fear is fear - if you have a fearful dog, consider the Bogeyman class to better understand how to address the issues of fear.

Exercise:

Place something in the training area that your dog wants (the proof), and be sure that your dog cannot get to it. If you have access to a second person, then have them hold it instead.

Before beginning, review what I have written above and re-watch the videos - make sure you reward generously for your dog's good choice to interact with you rather than the proof. The second step is to ask your dog to work a series of simple behaviors to earn a reward - Raika heeling in the second video is a nice example of this. You need to read your dog and decide what is appropriate. If you are just beginning proofing work, I'd suggest rewarding the dog for one single behavior - looking to you for direction. If your dog is more sophisticated, you might start with a simple reward, but you'd want to move up to asking for one or more additional behaviors before rewarding. Start small - we want success!

I'd like to see this exercise, if possible.

here is a brilliant example of handling a dog leaving to admire the treats - look how well Mel responds, and see how nicely Aria gets the idea - and starts to come to Mel faster and faster rather than hanging out with the treats: