What is shaping?

Shaping is the process of moving from an unskilled to a skilled performance through a series of closer and closer approximations of the final behaviour. Shaping behaviour is a very effective way of teaching complex actions. Any complex action or series of actions can be broken down into small steps, with those steps rewarded along the way.
In the beginning, almost anything goes as long as the dog is working with you and trying things. Try to reward any action that even suggests what you’re looking for. It’s like the childhood game of “warmer/colder” — you start with a broad base of acceptable actions, rewarding tiny steps and refining as you go along. As the dog gets closer and closer to the “target” behaviour, you no longer reward the same things you did in the beginning, but ask for more. Slowly chain the behaviours together as the performance progresses.

• Define the goal or training objective. Have a mental picture of the desired behaviour. What will you reward and what will you ignore?
• Design a plan. Break the training down into small steps. Organize and teach the steps in the most easily learned order.
• Start easy — reward early successes frequently. You’ll need to reward at a very high rate in the beginnning to keep your dog’s interest.
• Reinforce success and ignore failure.
• Teach each step individually.
• Gradually keep raising the criterion for success. Don’t get stuck at a low performance level.
• Teach each step well enough that it becomes a solid foundation for the next step.
• Be flexible. If it’s not working, try something different or go back to an easier step.

GO TO PLACE uses a bath mat, piece of carpet or a crate pad as a place for the dog to go and relax. It can be used to help define a DOWN STAY. Eventually you will be able to move the mat to different locations (visiting friends, for instance) and your pup will be able to come along for the visit without getting under foot. Learning PLACE is a chance to practice self control.

We will introduce a GO TO PLACE command as a clicker exercise. We will shape your puppy’s going to the mat on command, where he will lie down and stay there until released. The main steps for shaping this behaviour are:
1. go to the mat;
2. lie down on the mat;
3. stay on the mat.

GO TO THE MAT:

STEP 1: Drop the mat on the floor and watch your dog for any interest in it. It’s natural for him to turn to watch or even step forward to investigate it as you drop it. Click/Mark&reward any interest in it — even just a head turn. Pick up the rug and lay it back down, and again mark&reward any reaction: looking, touching with the nose, or stepping on it.
Reward by tossing your treat onto the mat — this gives you double the reinforcement value: the marker tells him going to the mat is what’s being reinforced, and actually getting the food on the mat again tells him what a great place it is to be!
STEP 2: Stand next to the rug and mark&reward any foot touching the rug. Lure the dog to the rug if necessary, but it’s much better to let him figure it out on his own. At this early step you need to keep your pup’s attention and interest, so help if you have to so you can reward frequently.
STEP 3: Put the food behind your back and lure or move the dog with your body pressure. Mark & reward when a front foot touches the rug.
STEP 4: Put the food behind your back and wait for the dog to figure out what he needs to do to get rewarded. Wait for that “A-HA!” moment as he looks at the rug “is this right?” and mark&treat the look. Reinforce his thinking and problem solving skills.
STEP 5: Now wait for the dog to move to the rug on his own before marking and treating. Begin to throw the cookie off the mat so he has to leave to go get it. This gives him a chance to go back to the mat for another click&cookie.
STEP 6: Wait for the dog to move to the rug on his own and stand on it with all four feet.
STEP 7: Move further away in small increments until you can send the dog from across the room. By now the dog should stand with all four feet on the rug and wait a few moments before you mark& reward.

LIE DOWN ON THE MAT:

STEP 8: Now that you can send your dog to the mat, go back to giving the reward on the mat. Give the treat down low, between the dog’s front legs, to help encourage him to lie down. He will probably figure out that lying down is just as easy as standing, and will lie down to get the rewards.
STEP 9: Add a voice command such as “GO PLACE” or “GO TO BED”. Mark&treat when he lies down.

STAY ON THE MAT:

STEP 10: Begin working on DURATION. You’re no longer going to click/mark&treat (marking ends the exercise, and we want him to stay on the mat, not think the exercise is over), but rather start giving treats continuously as long as the dog stays in a down on the mat. Say “good” (your “keep going” marker) occasionally as you give the treat.

When you’re rewarding movement and speed, give the reward away from the mat — click his feet touching it, then toss the cookie away. This gets him away for a chance to try it again.

When you’re rewarding duration — staying on the mat for longer and longer periods of time — give the reward on the mat. Be a Pez dispenser. Tell your puppy “good” and treat almost continuously. Now when you’re ready for him to get off the mat, use a release word such as OKAY and give the final treat off the mat.

STEP11: Move around the mat as you treat. Pivot in front from one side to the other. Take a step or two away and then return to treat. Continue to say “good” but start treating less and less, with more time in between treats.
If the dog leaves the mat, say “oops” (or whatever your “non-reward marker is), break eye contact, turn your back on the dog and wait for him to initiate going back to the mat on his own. If he doesn’t after a bit, you can give the command GO PLACE to send him back.
STEP 12: Go to rug, lie down, and wait to be released before receiving the reward. Use your release word OKAY every time you want him to leave the mat. Don’t let him decide he’s been there long enough.
STEP 13: Go to the rug, lie down, and stay for a variable length of time before being released. Sometimes release after just a few seconds; sometimes release after a minute or two. Continue to delay the reward for incrementally longer periods of time.
STEP 14: Move rug to different locations. Start with different rooms in the house before taking it to new houses, hotels or classrooms. Every time you go somewhere new, start easy — shorter duration, and more rewards.

The main drawback to luring is that the dog is focused on the food and not thinking about the task. When he’s just following the food, you’re doing the thinking for him, and he’s not really learning to problem-solve.

When a new exercise is introduced and the dog is lured into position, the food is visible and available. The dog stares at the food, and is not aware that he is being manipulated into a SIT. He passively follows the food, and somehow ends up in a sit, but he has not been taught to SIT.

Luring shows the dog that an action is possible, but it does not teach the action. The dog is not thinking when he’s following a food lure. He’s merely reacting: “Follow the food follow the food”. This leads to the common complaint “But he only does it when I have food”.

For learning to take place, the dog needs to understand that it’s his behaviour that causes the treat to appear. The dog must learn that there is a connection between his action and the food. If you lure for too long (many repetitions), the lure becomes part of the routine, and without it, the dog doesn’t perform.

Fading the lure.

Only lure with visible food for a few times — just enough to establish that the dog can perform the task. Then put the food out of sight — in your pocket or bait bag, or in a bowl on a table. You can continue to lure without food, using the same hand or body motions used previously with food. This is the physical cue to perform the behaviour.

Now our conditioned reward marker becomes important. As the dog follows the hand signal, mark “YES” (or click) and immediately reinforce. The treat becomes a reward after the dog performs a behaviour rather than a lure to cause the behaviour. The dog’s action causes the treat to appear; the treat does not cause the dog’s action.

Body language always overrides verbal.

Once the dog understands the physical cue and is responding correctly most of the time, it is time to “put the behaviour on cue” or name it.

Because body language always overrides verbal, if we say SIT at the same time we give the physical cue, the dog will only see the physical and not hear the verbal. We need to separate the verbal from the physical. Give the command, pause momentarily, then help with the hand signal.

Again, because body language always overrides verbal, if we say YES at the same time we reach for the reward, the dog will only see the physical movement. The thing he was doing at the moment he heard YES was watching the trainer’s hand reach for the bait pouch, so that’s the behaviour he thinks will be reinforced. The hand movements predict the reward — and he won’t be listening for YES.

You will know if your timing is bad because your dog will flick his eyes back and forth from your face to the bait pouch, waiting for the hand movement to tell him his treat is coming. You MUST say YES without moving your hand to present the treat or toy. YES … pause … then get the treat out and give it.

Antecedent –> Behavior –> Consequence

Correct: Command SIT –> Dog sits –> Dog is rewarded

Wrong: Show food, lure and say SIT –> Dog sits –> Dog is rewarded

One of the criticisms of “positive” dog training is the complaint “My dog only sits when he sees the treat!”

In learning theory, the antecedent is the signal that begins a behaviour and the consequence is the reward that reinforces the behaviour.

When we teach by “luring” the dog into a sit with a treat and then rewarding, we’re building the treat into the “picture” the dog has of the behaviour. The dog becomes dependent on the presence of food as the cue or signal to perform the behavior. Without the visible treat, the dog doesn’t understand the verbal cue “sit”.

For this reason, it is important to move away from luring with food as quickly as possible. And because “physical always overrides verbal”, if you want your dog to learn to respond to the command SIT, then the only thing that should cue the behaviour is the verbal command. If you give a hand signal at the same time as the command, the dog won’t hear (or learn) the command and will react only to the physical cue.

To correctly teach any new behaviour, start by luring the dog into position with food.  Once the dog reliably follows the food, get the food off your body (into a bowl on a table nearby, or into a bait bag) and lure with an empty hand. The luring becomes the hand signal for the behaviour. Make your physical motion less and less.

Begin to pair the motion with the verbal command. Give the command, pause briefly, and then lure the dog into position. Imagine the verbal command “hanging alone” for a moment, giving the dog a chance to hear it, before giving the physical signal. Continue to minimize the physical signal. Also be aware of any unintended help you may be giving the dog. Do you bob your head, lean over, raise and lower your shoulders, or reach for the treat bag as you give the command? All of these movements, even if they’re very slight, act as signals to the dog. The only antecedent to the behaviour should be the verbal command.

One of the things that really hit home this time is the difference between a terminal bridge word that means “yes, that’s exactly what I was looking for, here’s your reward” and a secondary “keep going” bridge that means “good, I like what you’re doing, keep doing it until I release you”.

I’ve been using “yes” and “good” for some time now, so it’s not the use that was my “ah HA” moment, but a refinement on how and when to use them.

Use “yes” as a release word. The dog should come to the reward in an active manner. So use “yes” when the dog is moving, and when you release, it should be active. If the dog is heeling and is in correct position with correct focus, then mark “yes!” and let the dog leap up to the reward. If you’re working on positions and your dog goes down quickly, mark “yes” and spring backwards allowing the dog to chase the reward.

Use “good” to build stability in exercises. The reward should come to the dog. If you’re working on “down”, use “good” and bring the reward to the dog, rewarding while the dog maintains the position. (Thus food is the reward of choice for stability exercises.) If your dog sits quickly on command in the moving sit exercise, and you’re working on duration rather than speed, mark “good” and walk back to the dog and give the reward in place.

Each are bridge words, but one is for stability, and one is for release.

Nonetheless, a marker word is far easier (at least for me!) to use than a clicker:  I always have my voice with me, and I don’t need an extra hand as I often do for the clicker. I use “YES” but there are potential problems with this word.

The word or sound you choose to use should be as unique as possible. This means if you choose to use the word “GOOD”, you can’t use “good” at other times. You can’t say “oh, you’re such a good dog” without diluting your mark. The more you casually use your mark, the more it is muted or dulled.

And the power of using a mark is its sharpness. With a clear, unique mark, you can manipulate your dog’s actions with almost surgical precision, teaching new behaviours without any physical force. (Why is this important? Because your dog will remember, and be more motivated to perform, behaviours that he was engaged in learning, rather than ones where he was a passive participant.)

For this reason, “YES” is not such a good choice, either. For example, while you’re training your friend asks “are you going to be at the trial this weekend?” You answer “yes” and now you have to “pay the dog”. Not only that, but you muddled up the power of “yes”. What was the dog doing at the moment he heard and understood “YES”?

Try to choose a simple but unique word, one that doesn’t come up often in conversation. “BINGO” is one several clients use… “BREAK” is another possibility; it could even be “MARK” (if you don’t have a Mark in the family!) or “CLICK” or even a tongue click. You could use a made up word. One client choose “SCOOBY” and that had a certain flair.

Whatever you choose to use, it should be:

• Short and quick to say
• Unique
• Easy to remember
• You can say it in much the same way each time

Finally, remember the three things a mark does:

• It tells the dog the thing he was doing the moment he heard the mark is the right thing;
• It predicts or promises a reward; the dog will never be cheated;
• And it ends the exercise.

(Read an interesting article “Clicker Bridging Stimulus Efficacy” that discusses learning differences between using a clicker and “good”.)

There are a plethora of methods for teaching articles, but I decided to start with them in the front yard off of the track.  I took out a handful of articles and scattered them in the grass.  Next came the dog and a bag of treats.  Of course, the dog looked at me, not the ground, because I have the treats!  So the treats have to go a little out of sight so I can show him the article on the ground.  Once he looks at my hand, I mark it, and treat on top of the article.  Stand back up, chances are pretty good he’ll look at the article just cuz there was food there a second ago, and you can mark and treat again.  Pick up the article and ask him to look for a new one.  Repeat above process for all the articles you’ve dropped in the yard.  Hey, this is EASY.

Next day I repeat as before, but while I’m treating at the article, I also give the hand signal for down (not a stretch, because this produces the same position he gets rewarded in for the down itself).  So while he’s down at the article, I keep bending over and paying him on the article.  For the first week or so, I let him mug my hand while I’m putting out the food.  More on that later.

Big mistake #1:  OK, it’s started, let’s see if it translates to the track.  I can tell you most emphatically, it does not!!! Poor boy was just really confused about why I was stopping him and asking him to down every 15 steps.  Repair:  Go back to articles in the yard for a while until they’re on command and highly reproducible, and he is actively searching for the next article once I pick one up.

Because I want to be able to straddle him while rewarding him at the article (something Ivan Balabanov recommends) I started doing this in the yard.   The first couple of times were OK, then he decided it was too much pressure, and as soon as I stepped over him, he scooted backwards and away.  This is probably a result of what I was trying to do on the track a few days earlier.  So… maybe too much to ask for just yet.

So over the last 5 days, I’ve gone out to the yard with a selection of articles.  For the last 2 days, I got all new ones by cutting up some scrap plywood and a chunk of alder into squares roughly 4″ across (10 cm).  These are easier for the dog to find (faster path to success) and can serve as a plate for him to eat from.

Overall, I think it’s working.  This morning, I know I am getting meaning to the command “find it”, because he puts his nose down and starts looking for it.  When he finds it, he stops and looks at me, and when I say “find it” again, he downs with the article between his front legs.  I even got to where I can straddle him briefly, but I moved from that position after a couple of rewards so as not to belabor it.  Also, I am now asking him to respect my hand a little more.  When he stops punching my hand with his nose, I release the food.  A little confusing at first, but it came along pretty quick.

So I see where we’re going, I think — smaller and new articles, only paying when he downs on the article.  The part I want to get next would be to have him down as soon as he finds it in the direction he’s facing when he finds it, rather than turning to face me.  I”ll try and put a post up when I get to that point.  Might have to refine that on the track.

Happy training!

Even after reading Clicker Bridging Stimulus Efficacy by Lindsay Wood, (clickertraining.com), which studied a group of untrained shelter dogs and found that the use of a clicker reduced the training time and number of required reinforcements 30% over the use of a verbal marker, I continued to think that clickers and markers were equal in strength.

But the other morning, the power of a clicker was really reinforced. I was cleaning the kitchen table and had finally gotten down to the Pleistocene era… at the bottom of the pile I found a couple of clickers. It’s been quite some time since I’ve actually used a clicker in training, but all the dogs’ foundation training begins with a clicker.

I picked up the clickers, and just like touching wet paint is irresistible, I couldn’t help but click one. Immediately I had seven dogs at my feet, each with the “what can I do?!” look on their face.

Would I have gotten the same reaction by saying “YES” distinctly? I don’t think so.