Thursday, August 8, 2013

THE SALVADOR DALI CAFE




Imagine what it might be like to live in a completely behaviourally random world.  You go to the cafe and order a latte, and the cashier asks for your hat and coat.  As a cooperative citizen, you give it to the lady who throws it in the trash, as she calls out “NEXT” and turns her attention to the person behind you in line.  You move down the counter to pick up your latte, and the barrista, comes out from behind the counter and grabs you and begins to waltz you around the Salvador Dali cafe and throws open the door and turfs you out onto the sidewalk.  On the sidewalk, a bear in a business suit offers you a Rolex, cheap from inside his waistcoat and when you say no, he pulls out some flowers and hands them to you and approaches someone else on the street.  Still wanting a latte you go back into the store, only to find that it is now filled with pink balloons and you cannot make your way to the counter.  After much struggling, you catch the eye of the cashier and mention your latte and she says “no latte today, m’dear, only champers and cheese” and hands you a plate of candied almonds.  Nothing you do can change the maelstrom of activity you have found yourself within.  How would you feel?


Living in our world as it is depends on understanding things like the rules of the game, that gravity controls what floats and what doesn't and who people are.  When things stop being predictable, then it becomes very difficult to learn what to do.  Salvador Dali had a talent for showing this concept visually in a playful and interesting way, never the less, few people would want to play checkers on a board where the pieces persist in falling through the board and where time expands and contracts independently of what the players do.   Image credit: rolffimages / 123RF Stock Photo

When we work with dogs, they are forever looking for the way through the maze; the way to control their environment, the rules of the game.  There are lots of games you can play and learn the rules through experience, and training is one of them.  Yesterday, I worked with a lovely young terrier who is trying to figure out the rules.  When she barks, she wants attention, and as a youngster, it worked well.  She can make her person do all sorts of things by barking.  Not wanting to disturb people, her person will come in to her and pick her up, or give her treats or tell her to be quiet or point a finger at her or tell her to lie down.  Like trying to order a latte in the Salvadore Dali cafe, everything changes at every step of the game.  So how can we make training an experiential fun game for the dog?


Think about games and activities that you have learned through experience.  If someone invites you to play Scrabble and you have never played before, it might run like this.  Your opponent will give you a hint or a clue or a starting point.  Perhaps they will start by giving you a tile tray and some tiles and ask you to make some words in your tile tray without showing the words that you have found.  Then your opponent will put down some tiles and the board and explain how the scoring works.  Then it is your turn and you can take your letters and place them on the board to make a word that intersects the first word.  You do this and your opponent scores your word for you and then it is his turn again.  He makes a word that intersects a word on the board and you work out the scoring together.  Turn by turn you learn the rules of the game, what strategies work and which strategies are ineffective.  This is exactly how dogs learn when we train using operant conditioning.


Coming back to my terrier friend (she really is a friendly dog who is just trying to make sense of her world), what we set up for training was a contingency that allowed her to learn some rules through experience.  The first thing we did was use a tether to limit where she could go; we set up a playing area so to speak.  Then we clicked and treated to remind her that the game was starting and give her some information about the game and how it would work.  Then we used four basic rules; if she was quiet, her person would stand close to her and wait.  If she was barking, her person would take one step back for each bark, until he got to the far end of the room.  After barking started, if she were quiet, her person would come back one step at a time.  If she lay down, her person would click and treat.  At first, she didn’t know the rules so she did a lot of different things to see what would happen.


When she barked, her person would step back.  This frustrated her and so she barked louder.  When her person was about twenty steps back, she stopped barking and he stepped forward.  Then she barked again and he stepped back.  For about five minutes, she learned how her behaviour affected the behaviour of her partner.  Then abruptly, she lay down.  Her partner came in and fed her a treat.  And she barked.  He backed up.  She stopped barking and stood up.  He stepped closer.  She lay down, he gave her a treat.  A very simple, but very predictable game, in which she controlled the behaviour of her partner.


This is how I think the best training works.  The trainer decides on the game for the day.  The trainer decides on what the contingencies are.  Then the trainer allows the dog to work out the contingencies.  The dog gets to decide if he wants to play or not and if the trainer has done his job well he has set up rules that make sense to the dog and the dog wants to play.  In order to do this we have to understand some things about the game and about the dog we are working with.


If I have a dog who LOVES liver, but doesn’t love cheese, then it doesn’t matter what I want the dog to do, cheese is not going to help him to learn the rules of the game.  Likewise if the penalty is something the dog doesn’t care about, then using that as a penalty isn’t going to work.  If in the scenario that I presented above the dog was afraid of the handler and barking made him go away, the dog would learn that barking resulted in something he wanted and he would do more barking.  I see this all the time in training.  The person thinks that their dog should like something, and they offer it as a reward, and the dog when working out the rules figures out that he will get something he doesn’t want to have if he does a particular behaviour.  This isn’t about the dog not wanting to play your game; this is about the dog not wanting what you have to offer. 


Sometimes the task we are asking the dog to do is of no interest to him.  If we ask a herding dog to go sit in a boat and retrieve ducks, no amount of liver is going to make that as fun for him as taking him out to herd sheep.  In the best training the activities we do with our dogs are of interest to the dog.  That said, there are always parts of the game that we might not enjoy; I hate setting up the board in Scrabble, but if I don’t do that part, I cannot do the part that I like doing.  Manners are an example of activities that your dog must do in order to get to do things he likes better.  Behaviours like greeting with four on the floor, taking treats gently, and keeping quiet are all behaviours that the dog must learn in order to be able to do things like playing agility, herding sheep or retrieving ducks.  Teaching good manners allows us to do more fun things with our dogs later and also allows us to establish that the dog doesn’t live in a random world where nothing he does affects the world he lives within.



Luring is a tool that I often see in the training game that can be quickly and dangerously misused.  Imagine if in my Salvador Dali coffee shop, the cashier kept holding out that latte that I wanted, but I could only get it by following her around.  I might tolerate a lot of Rolex selling bears and pink balloons to get my lattte, but I probably wouldn’t enjoy the experience a whole lot more.  Luring should be only used with caution and with respect for the dog.  When a novice trainer discovers that he can make the dog sit by holding the lure over the dog’s head, it is a short and dangerous step to using the lure to get the dog to do things he might consider otherwise risky.  Consider the dog who is not confident about getting onto a piece of agility equipment.  The trainer puts the lure on the equipment, and the dog is then in a conundrum; he can get the treat, but he isn’t learning to control his environment any longer.  He is conflicted because he wants the treat, but he has to do something that he considers dangerous to get that treat.  He doesn’t learn to play the game as much as he learns to balance the conflict between what he wants and what he doesn’t want to do.  Luring is even more dangerous when it is used to get the dog to interact with people or dogs he isn’t sure of, because he may at that point become aggressive.  Luring, properly used tells the dog how to position his body, but that is all it should be used for.  If the dog is concerned in any way about what you want him to do, pairing the thing he is concerned about to the thing he wants is a much safer bet than making his interaction the contingency that results in a treat. 

Luring is a tool that can lead the dog into the Salvador Dali Cafe.  It is best used to help the dog understand how his body ought to be positioned, but if it is abused and used to coerce the dog to do things that are uncomfortable or risky, then the dog loses control over how he interacts with the world and then he can end up in situations that don't make sense to him or that actually put him at risk.  In general, targeting can achieve the same results, with fewer risks.  Image credit: simsonne100 / 123RF Stock Photo


As soon as worry comes into the game, then it is not fun anymore.  Then it is time to play a different game; a game that will allow the learner to figure out that he is safe and that the world is a good place to interact with.  When the dog is worried the game should simply be “see that scary thing?  It produces treats”  This game is great when it is played with the frightening thing far enough away that the dog can cope with his fear.  Only when fear dissolves can you switch back to a game of “if you interact with the thing you are worried about, you will get a treat”.


When we think about training as a game, we can set up a series of rules and outcomes that the dog can be successful at.  Teaching dogs to be confident is contingent on the dog being repeatedly successful.  If the dog is successful over and over again, the dog starts to think that he can do many more things than he used to be able to do.  Perhaps the worst thing we can do to our dogs is create a random world where they cannot control what they live with.  Overwhelming dogs with repeated conflicts or failures results in a dog who lacks confidence and who doesn’t want to participate in training games.  If your dog feels like he is living in the Salvador Dali coffee shop, don’t be surprised if he stops participating in the training game, or if he becomes tense or fearful or anxious.  Just writing about the Salvador Dali cafe is difficult because none of it makes sense and there is no control over the outcomes.  When there is no control over your world, learning is inefficient and upsetting and pretty soon you have a dog who just doesn’t want to participate any more.


1 comment:

  1. Interesting perspective. I was thinking about this post the other day when we met up with a neighbour's dog. This dog is very friendly but gets insane with excitement when meeting another dog.

    From my dog's point of view:

    We're walking towards another dog. Oh goody, we're stopping to say "hi" - let me sniff this dog's rear. All the owners seem happy. This other dog is spinning at the end of his leash - game time? I'm in! Jump and pounce 'R' us! Choke - Jen's pulling back on my leash - ack. What happened? Oh good - now the other guy is telling me what a good dog I am. He's giving me a biscuit! Oh no. Now he's mad at his dog for some reason. But wait - he's talking to Jen again and we're back to playtime . All together now - POUNCE! Oops. We're stuck together with tangled leashes. "VINCENT! :(" "RADAR! :(" What did I do? Let me try sitting. Oh good, that worked. Note to self - sit when Jen yells.

    Hmmmm.....

    ReplyDelete