Friday, June 13, 2014

How to create games that teach a behavior

How to create games that teach a behavior.

Step 1: Thoroughly examine the behavior you want to teach.

What is the last action the dog does; what is the first; determine all the different actions the dog would do from start to finish.

For example: teaching a dog to skateboard.

I’m using a skateboard here mostly because it’s not an easy behavior to teach due to the movement of the board and because Micah taught it to himself with very little input from me. I had the pleasure of watching a dog who knows how to think and how to create and figure it out for himself.
1.       Notice the skateboard
2.       Examine the skate board (silly thing you say, but what I've noticed every time Micah gets on the skateboard he has to sniff it first. It may be a really quick drive by type sniff, but he does it every time)
3.       Put a paw on the board
4.       Put another paw on the board.
5.       Put a third paw on the board
6.       Put the fourth paw on the board (again, with Micah I notice he puts all four up, then takes one or two down to paddle)
7.       Change posture to face forward
8.       Change posture to lean forward
9.       Put one, maybe two feet down and push.
Those are the major steps.

Step 2: Examine the different actions involved in the behavior and look for differences and similarities.

The reason for looking for differences and similarities is that you may be able to lump actions together into a single game, or you’ll need to create a game for each action.  Sometimes you may even need to break an action down into even smaller pieces.
Taking our example of teaching a dog to skateboard, what we see with 3, 4, 5 and 6 are what appear to be similar actions – put a paw on the board.  But if you look at it closer, what is actually happening is that you are “adding” paws, not just putting a single paw on the board in each step.  It’s usually fairly straight forward for a dog to put the two front paws up on a platform with little difficulty, the problems start with the rear paws.  Many dogs do not understand that they can move their rear paws independently and some don’t even realize they have rear paws and legs.  The rear often just follows along and the dog often looks clumsy when maneuvering around obstacles.
So, we’ve examined our actions and realize that our dog may have difficulty with one or more steps.  You could create more than one game for the difficult actions so that the dog learns about that action from different viewpoints or you could break those actions down farther and create two or three games to facilitate the dog’s learning curve.
It’s better to have more games than less with the understanding that in many cases, the dog figures out the final game without having to do all the intermediate games.  Many of the games necessary for teaching a behavior may also not be necessary because the actions were learned as part of other behaviors and just need a refresher or associating a specific game with a new piece of equipment, obstacle or environment.
Micah learned to skateboard in less than 10 minutes.  He already understood about platforms, had decent rear end awareness, wasn’t afraid of moving objects and really, just had to realize that 1) at least one foot needed to be down to push, and 2) that this object could move while he was on it in a different manner then a teeter.  I did nothing.

Step 3: Look for already created and tested games that can teach any of the steps in your target behavior.

There are many games already out there on the Internet, in books and on DVD.  There are also many variations of the more well-known games like “go to place”.  As you are looking, keep in mind that even though a game seems to be specific for another behavior, games can have multiple uses.  For instance in the classes I teach each has 48 games (8 weeks of 6 games each).  Many of those games are repeated in several classes. Crate Games, “It’s Yer Choice” and 1-2-3-Break from Susan Garrett’s Ruff Luv book are in every single class.  These three games teach self-control and the ability to think through arousal.  1-2-3-Break and Crate Games also teach an awesome stay. “It’s Yer Choice” expanded can resolve counter surfing, eating poop and stealing popsicles from toddlers.
So for our example of teaching a dog to skateboard we could use any version of targeting for steps 1 through 6 and 9.  There are several games from the agility world for teaching stopped contacts that we could use for getting the dog to put all four paws on the board.  We could utilize other forms of platforms to teach the fundamentals of “paws up”, what to do on a platform and how to use one.  “Look At That” from Leslie McDevitt’s Control Unleashed could be used for step 1. My “What’s That” Game for step 2. My “Dig it” or “Tap Tap” games for step 9.
The most important part of looking for already created games is to make sure they can flow into the games that will follow them and from the games that precede them.  “Look At That” flows right into “What’s That” and What’s That flows into any variation of “Touch That” (targeting) which then could potentially flow into “Tap Tap” for the paddling once all four feet have made contact with the board.  All five of those games are in reality just variations of targeting with eyes, nose, and toes.
If your dog already knows the games you select, and most of the steps for your new behavior have been covered with known games, as with Micah and his skateboard, teaching the new behavior is a breeze. But if you are new to training, or new to structured game training, you will have to teach each game.

Step 4: Create a game

“Dog training should be a TWO way flow of information. We don’t lure dogs with treats or physically manipulate them to the place where we want them to be; We create challenges, in the form of games with obvious solutions, then we sit back and watch them make their choices.”  Susan Garrett
This is actually the easy part, but there are a couple of rules about structured games that should be followed.
1.       Remember that a game does not control the dog or the human.  The human controls the environment so that the game is easily apparent to another human and the dog, and the dog makes choices based on that environment.
2.       Reinforcement is given for choices, choices that lead to the behavior the game is designed for.  There should, ideally, be only two choices in any game a) do it and b) do something else. Reinforcement is only give when the dog chooses “a”.
3.       There is no correction, punishment, redirection or other “handling” by the human necessary to play the game.  If you’ve designed the game so that you are constantly feeling like you need to “help” your dog, then the games has a) too many choices and needs to be split into smaller games or b) you have not sufficiently set up the environment so that the “do it” choice and the “do something else” choices are clear or c) the proper foundation games have not been played.
4.       Each game should take no more than 3 minutes for the dog to become 80% proficient at playing.  The original choice should not take more than a minute for your dog to figure out, any longer than that means you should probably split the game.
Now comes the creation part.
1.       Pick one of the behaviors that doesn’t already have a game. In this case we’ll pick “face forward”.
2.       Decide where your starting point is and what the end point should be.  For “face forward” the starting point would be either looking down at the feet, at you or at one side.  The end point would be facing forward without the feet leaving the board.
3.       Figure out what the environment would have to look like to encourage the dog to go from the starting point to the end point. With “face forward” it would have to be something to encourage the dog to face forward, but keep her in place on the board.  There are a couple of things that could be done here.  If your dog had some self-control and understood a release word, you could put a toy a foot or so in front of the board and wait for your dog to look at it.  Then release and reward with the toy.  If your dog doesn’t have a lot of self-control or doesn’t understand a release word yet, you could ask for a hand target or just wait til your dog looks forward and then reward.
Remember that you are only manipulating the environment.  Luring with a treat is manipulating the dog.  The example I gave for “face forward” creates a static behavior.  The dog stands still and looks forward.  But our final behavior is going to be a dog who is moving forward on a skateboard.  The static behavior would work, which is why I include “lean forward” as the next step, but you could also combine the two with a simple run up to the board, get on the board, briefly stop and then run off in a straight line.  This would encourage the dog to move with the skateboard once you’ve taken any anchors off.
As you can see, there are only three steps here.  The game should never be complicated; it should be a simple start, end.

Step 5: Discover what the reinforcers for this behavior could be. By definition, reinforcers both depend on behaviors and sustain them.

Remember that all games must have a consequence and for our purposes because we are trying to teach new behaviors or increase already known behaviors, those consequences need to be positive.
Things that seem like rewards sometimes aren’t: what matters is that the behavior is learned or increased, not the intention of the teacher.  If a “reward” has no effect on a behavior, then it’s not a reinforcer.
Even with these short games, you can run into boredom. “Boredom” indicates a lack of reinforcers, not a lack of stimulation.   Boredom can happen even with high value reinforcement especially of the food variety.  Satiation can occur and then your liver or cheese is no longer reinforcing.   So always have more then one TYPE of reinforcement, not just more than one flavor of food.
You must also take into consideration the behavior you are teaching.  If your behavior is a moving behavior, your reinforcers should be of the same type – chasing toys, your hand or food. If a static behavior, then tug or leaping targets would make great reinforcement – anything that keeps the dog relatively still but doesn’t create satiation or boredom.
When choosing things to be reinforcers, you best choice are those things and activities that your dog already enjoys.  Chasing squirrels may be a great choice as it is highly reinforcing, but it is impractical.  However, chasing a “squirrel” at the end of a flirt pole is highly rewarding and highly practical.  So just as in the creationg of games, think outside the box with your rewards and reinforcement.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.