And now, back to our regularly scheduled programs…
Reliability. It is what we are all aiming for in our dog training. Sure my dog can do some crazy behavior, but can it be done reliably is the big question. There are many things my dogs can do that they cannot do reliably. Rio can blow bubbles in water, but not reliably. Shayne can do a handstand, but not reliably. Neither of these tricks are terribly important for anything other than showing off or having fun so reliability isn’t all that crucial to me at this point. Reliability for important behaviors is what everyone wants as an end result but it takes time and work to get there. Unfortunately, I see many dogs who do not have generalized reliability with important behaviors and handlers are getting frustrated with their “stubborn” dog.
Sure, Rex walks GREAT on leash… when we are walking in our driveway at nighttime or when at class but ask him to walk down the neighborhood sidewalk on a Saturday morning and he’s like an untamed beast. It’s really important that people work towards reliability with whatever behaviors they find critical for daily life. Reliability does not happen over night and it does not happen without some focused work.
One of the most common errors I see novice handlers (and even some experienced trainers) make is that they go way too far too fast. Instead of building on reliability, handlers will push forward on a behavior that isn’t yet reliable at a lower level. They may remove treats too early or they may increase the criteria before the easier criteria is reliable.
I like to equate it to taking off the training wheels on a bicycle when a child has not yet mastered riding WITH the training-wheels attached. If a child is falling off of the bike WITH training wheels, it is pretty likely that they will struggle riding a bike without them. Will they eventually figure it out, probably, but after how many broken bones, lacerated foreheads, and outbursts of tears because of the pain. Why go the harder (and more painful) route when an option exists to master approximations of the end behavior until the end behavior is accomplished?
Many of my students really focus on “getting rid of the treats” and want to do this ASAP and I try to tell them that they should keep the treats until they LOVE the behavior. Until a behavior is 99% reliable WITH food rewards (in a given environment), I do not remove the food rewards (*or a very high level reward be it a toy or food). I need to make sure the behavior is very strong WITH the reinforcer before removing the high value reward from the mix. If the behavior is weak before removing the food rewards, it will likely continue to weaken and eventually fall apart completely. I see this all the time when handlers start working a behavior outside. They start using treats but the dog is still pulling regularly although the dog is more under control and pulling less than it was prior to classes. Even though the dog isn’t really reliable at this point, the handler sees some progress and they drop the treats cold turkey. It never fails that the dog’s behavior quickly falls back to where it had been previously.
Last weekend when I was at the agility trial, Rio walked at a beautiful loose leash the entire time (whether he had the full 6ft leash or only a foot or two). There were dogs running in the ring, dogs barking at him, dogs passing right next to him, other dogs tugging right next to him, people walking right next to him, all sorts of crazy distractions and through the whole event I didn’t use any food to reinforce his LLW. The reliability for this important behavior had gone through enough proofing and building up that it was easy-peasy for him.
But I didn’t go straight from practicing LLWing in a classroom with treats to LLWing at the agility competition with no treats. We spent time building the behavior in all sorts of environments using various rates of reinforcement. Once he was reliable in a more typical outdoor environment with a low rate of reinforcement, we pushed the envelope by going to a busy outdoor outlet mall (thousands and thousands of people shopping) quite a few times to train in extremely distracting environments. The first few times I went loaded with treats to have a high rate of reinforcement around all the people, kids, strollers, other dogs, food, and going in a few dog friendly stores. It was quite literally a reward every 5 or 6 steps at the first little bit during our first trip. As he grew reliable with the treats, I lowered the rate of reinforcement as we went along. By our last trip to the mall, I had a handful of kibble in my pocket for the whole day, just in case.
Other than convenience, I’m not sure what the focus on getting rid of treats is all about. It’s such a driving force that people do it too early and ultimately sabotage all the training they had done prior to dropping the food rewards. It’s not as though it takes years to get reliable behaviors for what most average pet owners are looking for (loose leash walking at a park/neighborhood, not jumping on people, listening at home). Of those typical desired behaviors, LLW is probably the most sought after and the most challenging for people but with focused work, it’s not hard to get there. I recall with former foster dog Chase, he went from needing 2 full string cheeses to just a few pieces of kibble within 2 weeks to get through a walk without being unruly on leash. This was a long neighborhood walk seeing squirrels, chipmunks, other dogs on walks, dogs behind fences, kids on bikes etc. We used lots and lots of treats to build the behavior (starting in the backyard) and then as he was reliable with treats, they slowly went away until I carried some kibble “just in case”.
Moral of the story, build reliability before reducing your rewards. Once you have a reliable behaviors with rewards you can methodically reduce the rewards (if done well, this can be done pretty quickly actually, especially if you start using real-life rewards).