Generalizing What a Dog Learns During Training
Theory is what I call the things he learns in "formal" training sessions. The place where he's set up to succeed (in the beginning anyway), has the concept broken down into smaller pieces and tasks, and basic mastery takes place.
Application - well that's probably obvious, but it's what I call taking all those things learned above and applying them to actual tasks and increasing the challenge as time goes on and ability improves.
An Example - Learning Targeting
When I was teaching Wally to target things, I took him to the "classroom" and used objects that would attract his attention naturally and I put them almost in his face making it impossible for him to miss with any forward motion to the object. Then I had something to mark and reward, giving him feedback. The short distances also let the pace pick up quickly, keeping him interested and alert.
As we progressed, I would put the object in a more natural position. By "natural" I mean more like how objects will be placed in the real world. On tables, on the floor/ground, not always right in front of him, different shapes, sizes, and colors, and my hands as well. This was all still in the "classroom" so he didn't have too far to go and minimal distractions (or in this case most of the "distractions" would be other objects he could target).
The Classroom - Where Learning Takes Place
The classroom is where we do our training. It's an area that's low distraction, no unusual sounds going on and things are very structured and focused on a certain aspect or part of a task. Or, in the case of shaping, we're working on getting on behavior and then mastering putting it all together with consistency and fluency. This is where the cues get mastered as well if we're at that point.
Everything here is set up to keep Wally interested and focused. It minimizes him "playing around" or looking for things to offer behaviors on, distracting his brain and causing me to do too much redirection, and all things we need are right there on hand - any objects, the treats, the clicker if I'm using one, anything we might need.
Like any teacher, you have to have control over the classroom, but you don't want to interfere with the student's learning. You also don't want to be boring - so I try to keep things moving fast, the reinforcement is delivered fast and he can consume it fast, and there's a controlled excitement. I keep my canine student focused, but I only redirect if he's totally lost or getting sidetracked for too long.
And of course, tailor the environment to your student. This is the testing ground, so play around with it. Move the object closer, try a smaller piece, make the answer as obvious as possible without telling him what it is.
The classroom should be a safe place. I had to make sure I didn't scare hm, which was difficult sometimes in the earlier days, and I had to make sure that my feedback wasn't necessarily always positive, but always delivered in a way that I don't turn off his brain and turn on his appeasement. I didn't want apologies, I wanted an enthusiastic next attempt!
Time for Recess...and an opportunity to transition
Recess is important. This is where I'll take what we're working on and make a game out of it. We'll leave the classroom and usually go outside, but still in a relatively distraction-free environment. At this point, we'll play a game involving what we're learning.
Of course, I love games as part of training anyway, and this ties right into it for me. I'll pick a fun game that I know he likes to do, and make starting the game and it's continuing based on him doing what we've been working on.
This in my mind is also a transition from "theory" to "application" as it's using the concept in a real way but not quite with all the challenges that would accompany a full real world use. It is still also a learning period so mistakes are overlooked and things don't have to be real tight yet.
For a behavior like targeting, he has to touch an object with his nose, then we'll play a bit of a game. Or if we're targeting my hands, I'll move them around and play a bit of "tag" with them where he has to put his nose on my hands and I'm moving them around, trying to get away from his nose. For other behaviors, for example, lying down, he'll get a game once he lies down and then I'll stop and he only gets an other round once he lies down again.
Of course, he needs a break to rest his brain and do some doggy things. So he gets time to just sniff around and explore and just enjoy some time to unwind. If the weather isn't favorable, I'll just play some fun silly thing with his toys or such after the structured game.
The Real World Curriculum
At this point, the behavior is as good as it's going to get. The classroom is no challenge anymore, it's time to apply it to real world situations and various tasks that he'll be asked to do from this point on. Still, though, challenge is still controlled as much as possible, but, as we know, real life can't be scripted like the classroom.
What's your approach?
How do you approach training and generalization?
In this aspect, it's taking a classroom behavior and applying it to something he'll be asked to do on cue, such as opening and closing doors. To use the targeting example, touching my hand got transferred to touching the door. And then we would build on it from there to pushing the door and keeping it up until it's closed.
As with situations training, it will be a challenge at first as the dog has to transfer his knowledge to the new thing being asked of him. The reinforcement once again came quickly for any success and the challenge was kept in check. The advantage to this aspect of real world training is that things can still be scripted to some degree. The door can be easier to close, he can be allowed to start closer to it, etc.
This allows him to build on his knowledge and helps keeps behaviors fresh in his mind. It also gives him ways to help him navigate the world. Wally can get around the house on his own since he can open doors. He can help me as he closes cabinet doors on cue, allowing me to continue cooking or such. It gives us ways to work together.
With each opened and closed door, he sees the behavior can work for him. This is it's own positive reinforcement as he can go to the next room, or follow me (or find me) and things that were once impassible obstacles are now just things to be pushed out of his way. Also, task training can snowball and teach him problem solving. Next time something is in his way, he may well just push it out of the way. Since nose touching worked so often, he will use the behavior in other situations to see what happens.
One aspect to learning how to perform behaviors in the real world involves diverse situations and distractions. Here, Wally had to learn to use (and even remember) the behavior in situations that he or I can't predict and could change at any time. This doesn't have to be anything dramatic or anything, even something as simple as a dog barking out the window or kids playing or even a sudden gust of wind changes the situation and is one more thing that wasn't experienced in the classroom setting.
Of course, his mind will wonder, and in the early stages, it is still a lot like being in the classroom but the challenge will be higher. I had to make sure to keep him upbeat and encouraged and the reinforcement much like in the early stages - fast, frequent, and the task was still on the easy side. The first steps are primarily to get him confident. "I can do this!" I wanted him to think.
Over time, the reinforcement became less frequent and the challenge increased. I would allow less mind-wandering, required tighter responses, and allowance for surprise distractions decreased. Wally was to maintain the behavior even if a kid suddenly screamed or a squirrel (or cat) was close by.
Being creative can really help with some situations. Sometimes, I would click and let him chase the squirrel, for example. To use the targeting example, he would touch my hand or leg, or whatever, get the click and then I'll point out a squirrel and let him chase it.
Real World and Classroom are Partners in Learning
Combining scripted situations of the classroom with the aspects of real world training can rapidly cement behaviors in his mind. It helps to generalize behaviors, meaning that the behavior means the same thing in any situation and environment, and it can help with driving a behavior to being a self-rewarding behavior, helping making reinforcing the behavior easier and making it more likely that the dog will perform on cue with less delay.
This also means not to be afraid to go back to the classroom to help him if he struggles. Try to recreate the situation or task as much as possible in the classroom as well as refresh in his mind what he is being expected to do, much like practice for an athlete tries to recreate the actual game situations he or she is expecting to face.
Also, while training and learning never really stops, always try to give the dog time to rest and unwind. I always let Wally unwind and sleep once we are done with our "school day" even though we are always training in some regard. Remember, sleep isn't just rest time, it's also allowing him to absorb the information in his mind.
© 2011 Brian McDowell