Go Faster! More Cookies!
I spent part of the weekend trying to find a link to a study that I read years ago that looked at the skills/mechanics trainers used more effectively than average pet owners. Unfortunately I couldn’t find anything–so I can’t cite the research–but one of the features it talked about what that trainers had a much higher rate of reinforcement than the pet parents. I won’t give you the number because I’m not really sure I recall it but it was something drastically different (like 6:1–so for every 6 reinforcements a trainer offers, a pet parent will offer 1).
I have to say that it doesn’t shock me at all that trainers have higher rates of reinforcement and that the rate of reinforcement effects how a dog learns new skills. It’s certainly one of the things my students struggle with. There is this idea that dogs should respond just because they want to please us and that using food treats is bribery and I think many pet parents really struggle to let go of these ideas even when provided with ample information to their contrary. There is also an incredible amount of stigma attached to using treats in training from non-dog folks or old-school training folks. I’ve had more than one Joe Schmoe laugh at me when I’m training with a high rate of reinforcement or make some snarky ignorant comment. I have to remind myself that my dog will be walking beautifully when I’m done and they will still probably be pulled around or yelling at their dog to heel aimlessly! The social pressures that people feel to stop using treats is something that really interests me on one level, while also totally confounding me–what is so inherently wrong with rewarding dogs with food for good behavior? Though, I suppose, this conversation is for a different blog post!
When handlers can get past the social pressures and focus on the efficacy of their training, they can really start to improve their training skills by maximizing their rate of reinforcement. A higher rate of reinforcement provides a dog with a large amount of feedback on their behavior this feed back makes it easy for the dog to figure out what works. It also, in many situations, prevents the dog from making wrong choices early on because the reinforcement is flowing in so quickly that there isn’t much time to be wrong. Depending on what skills are being practiced and reinforced, a high rate of reinforcement often translates into many repetitions which also contributes to learning quickly.
Using a high rate of reinforcement really allows you to teach new skills very quickly (for the aforementioned reasons). Since you get the desired behavior quickly, it actually allows you to start slowing down the rate of reinforcement pretty early on while still maintaining the reliability of the behavior. Most people tend to get stuck just teaching the dog the behavior so it takes longer for the dog to acquire the new skill and until they are reliable with the new skill, you can’t reduce the rate of reinforcement and expect to keep the behavior reliable.
Anyone who has taken my group classes knows that the method I use to teach loose leash walking involves using a very high rate of reinforcement to start. We start with a step-treat, step-treat, step-treat, step-treat pattern–I start by having the handler take backwards steps (with the dog in front of them) to help the handler learn the fast rate of reinforcement since the dog is less likely to pull if the handler is fumbling to pay out when they are moving backwards. It doesn’t take long for the dog to be following the handler closely step by step wherever she moves going backwards and we transition to going forward. The dogs tend to take this transition in direction easily and don’t even trying to pull because the treats are coming so quickly right next to the person. The result is that we get a dog loose leash walking reliably very quickly (for a high rate of reinforcement). Once we have the desired behavior in a specific environment, we can reduce our rate of reinforcement. The way that I go about this is to introduce a figure-8 pattern–I start by instructing students to reward the dog 5 times on the very short figure-8. This translates to a treat every other step or so. When they have mastered the figure-8 with 5 rewards, we drop the treats quickly down to 3. In one class the dog will learn to walk forwards for a step-treat, step-treat, step-treat all the way down to going 6, 7, or 8 steps in between treats.
The high rate of reinforcement to start allows us to drop the rate of reinforcement down quickly without the behavior declining because the dog knows what pays out. Trigger, an SJC day-training graduate, started out being a beast on the leash. He would pull like a tank and was pretty wild. During our very first lesson I started him with the backwards step-treat, step-treat pattern and by the end of that first lesson we were doing figure-8s in front of his house. By the third visit he was walking up and down his block (in an urban environment) reliably and by the fourth visit he was able to walk on a major city street without pulling–with a high rate of reinforcement. By the end of our time together (a total of 12 visits where we taught all the foundation behaviors, not just loose leash walking) he was walking around his block with treats only once per block and could walk on the very busy city street with 1-3 treats per block (depending on how busy it was, if there was a food restaurant on the block, etc).
You have to make sure that you are using TINY treats or making the pup earn their daily meals as rewards to make sure you aren’t over feeding the dog because with a high rate of reinforcement you will be giving tons and tons of rewards. I pay as cheap as possible because I know I will be doling out lots of rewards and I don’t want my dog to overeat. For example, if I’m using Zukes training treats, I will break a single treat into at least 6 pieces and if I’m thinking ahead, i can cut them into 10 or 12 rewards out of a single pea-sized treat.
Take your training to the next level by incorporating high rates of reinforcement and you will see how much easier it is to teach the new skill and to reduce the treats than if you have a slower rate of reinforcement in the early stages.