TO SHAPE OR BREAK YOUR DOG. IS
THERE A QUESTION?
by Ed Bailey
Printed with permission
First published Gun Dog Magazine 1991
Sometime back, about forty or fifty years ago the learning psychologists were promoting the idea of nurture. All animals including humans arrived fresh out of birth canal or egg, as the case may be, as clean slates just waiting for life to indelibly write a series of messages which determined how the animal did everything. They thought all behaviour was learned. Therefore, with proper procedures, an animal could learn to do anything if properly exposed to the learning process.
This properly done teaching was called shaping. The chief proponent of shaping was B.F. Skinner. With his techniques pigeons could be taught (and were) to play ping pong or to fight when a green light was turned on or to be peaceful when the white light was on. Rats were taught to climb a ladder to a suspended cable car, get in the car and pull themselves across an open space to a platform, go down from the platform to another car, go through a maze and end up getting a piece of food for all their work. To those of us who can’t chew gum because we bite ourselves even while sitting down, the rat’s learning sequence is phenomenal.
But what does this have to do with the real world of hunting dogs? Shaping is the psychologists’ jargon for teaching an animal to perform a set task by giving it positive reinforcement for each correctly performed increment of learning. Positive reinforcement can be anything the animal likes, ice cream, a pat on the head, a nice word, anything pleasing to it.
The increment in leaning can be very small, but only progress is rewarded and mistakes ignored. For example, to teach our dog to sit (I know no real pointer man would have his dog sit, which is too bad because it is a good way to maintain control, something retriever people have always known). Back to sit. You, the trainer, tell the dog “SIT” in a normal voice. The dog doesn’t know sit from shoe polish but because he is standing beside you with a leash snapped to his collar, he is a bit intimidated and lowers his rear end, maybe only a hair, and has dropped his tail. Now reward with a good boy pat on the shoulder or scratch his ears, encouraging words of praise or a food reward. Repeat SIT, and wait until his tail is folded under and he lowers his rump a bit more than the first time, then reward again. Then repeat the SIT again and wait for a bit more lowering of the rump before rewarding. Each of these repeats is considered a trial. It might take many trails 30 or more, or a few, three or four before the dogs sits quickly on command. What the dog has learned is to associate the SIT with the reinforcement (reward) and the only way to get reinforcement is to plunk his butt down. He still doesn’t know sit from shoe polish; it only means that to get the good old boy routine he must perform a certain movement pattern when asked to.
There is no need to pull up on the collar for straight fingered pressure on the rump. The sit is a natural, normal movement pattern; all dogs do it. What is learned is to do it when told, when the magic sound SIT comes from his good buddy, the dog associates a stimulus, the word SIT, with the response sitting down because of the retroactive influence of the reinforcement or reward. The reinforcement must always be positive, must be something the dog wants and is happy when it gets it.
Reinforcement can be anything but cannot be negative. The reinforcement must follow quickly any step in the right direction no matter how small that step might be. And sometimes the steps toward the end point established by the handler are painfully small. But only correct responses are to be reinforced, never incorrect even if only half wrong. If the objective is to get the dog from some point A to a miles-away point B, a half-inch toward B is much preferred over a half-mile back toward A. Toward B is rewarded; toward A is ignored. Punishment is not physical; withholding the reward is the punishment the dog will get.
An additional twist tot the shaping can also be inserted here. Let’s say our dog is to sit by our left foot facing in the same direction you are. You are lining him up to retrieve the second bird or the double you have just made on quail and you want his undivided attention. Hold the reward treat in your left hand between thumb and forefinger so our dog can see it. Give the sit command in a normal voice (dogs have pretty acute hearing so a yell is really not needed) and move our hand up and back over the dog’ head. Your dog will sit automatically in his attempt to keep his eyes glued on the cheese bit or dog treat, whatever you use.
Lower your hand in front of the dog to mouth level and give him his reinforcement.
The dog’s brain is not only associating the word sit with the response, he is also associating the hand movement with the response. After some quick repeats so your dog responds quickly and correctly on, say, 9 out of 10 trials, use only the hand movement without the word and use the word without the hand movement.
Vary the schedule of reinforcing the dog so he doesn’t know when the reinforcement will come, but he is gambling that it’s the next time. Your dog has two commands for sitting. Either the spoken sit or the hand movement will mean put your butt down. Occasional reinforcement will keep him sharp and focused on what you want.
Breaking is almost the antithesis of shaping. The concept comes from the Wild West when the cowboys corralled wild horses, selected one, put on a nose noose and saddle and tried to stay on ‘til the horse gave up. If the horse didn’t break a leg or its back (or the cowboys leg or back” it was added to the string. The horse did it or it was subjected to enough pain to force him to do it.
The concept of breaking a dog sprang from the same root. Reinforcement is negative. An example of breaking to wing and shot comes to mind. The way it is done is to plant birds by dizzying them or putting them in some sort of release cage in the field or if enough birds occur naturally, use wild birds, but this is rare because it is not practical in terms of time and control. The dog is allowed to approach with favourable wind but either dragging or being led by a long check cord (the word “check” is the operational one). When the dog scents game it might point, hopefully, or it might charge to flush. Either way, the handler or dog flushes the bird and the dog chases.
Just before reaching the end of the check cord the yell of “WHOA” echoes off yonder mountainside. The handler hauls back hard on the line, the dog spins around the air, falls with a crack on his back and is quickly stacked back up where the bird was with another WHOA bellowed out. The clever dog after doing this a dozen times on a 30ft. line runs out 29 feet, does a neat back flip, darts back to where the bird was and stands staunchly, smirking while the bird flies away.
The normal dog, which thankfully most are, gradually figures out to chase is to get hurt and so stops chasing a flushed bird. All reinforcement is negative but is effective.
In today’s world of high tech dog training we can use the shock collar instead of the check cord. The concept is the same, only the effective distance is different. The learning involved in breaking follows the classical condition sequence. The stimulus word, Whoa, is followed closely by the negative reinforcement (the rope yank or the electronic yank) followed by the response of stopping. All of this training is dependent on the dog being in full chase. True force retrieving, whoa, down, sit are all trained behaviors in which the idea of breaking is used. However, shaping techniques work as well or better, but not quicker.
To some degree the payoffs of shaping and breaking are similar but there are some major differences. Shaping is usually slower and requires more handler patience. The professional trainer to whom time is money usually favors breaking. But the dog might be an automation. It does exactly what the command says with only thinking of maintaining life and limb. If the dog goes to a soft handler who doesn’t or can’t use strong negative reinforcement when called for, the dog can soon revert. Dog does just the minimum expected and less if he can get away with it without being clobbered, stomped, yanked, shocked, chained or some other painful reminder. Most people are not willing or capable of the heavy-handed approach. And then it’s back to the pro for refresher courses.
The shaped dog that has arrived a the same level of performance (obedience) as the broke dog keeps performing well if the positive reinforcement comes occasionally. Much like the lottery player, the dog is gambling that the next time he does the sequence correctly he will be a winner and get his ice cream or good old boy pat. So performance decrement is less with shaping than with breaking.
Another payoff is more subtle than loss of performance arises from the fact that the dog has gone through learning steps to arrive at the end point, he has added 2 and 2 and comes up with 5 or even 6. The broke dog got to the end too, but actually started there so there was no building on acquired knowledge. The dog that has learned has not only learned the task, he has learned that learning paid off and he has learned how to learn. This is the dog that seems to understand every word you say, solve problems, figure things out, size up the situation and make things happen. This is also the dog that is better able to concentrate and figure out a difficult track of a wounded bird. The shaped dog is more apt to “think” because it has learned to think. While the exceptional broke dog will show brilliance, most will have only the occasional flash of thinking.
With shaping you have made the dog believe he wanted to do it; with breaking you have convinced the dog it is easier than being hurt.
No training is entirely one way or the other. Judicious punishment is essential in shaping the easiest-to-train dog. Proceeding one step at a time should also be an integral part of the faster breaking methods, particularly with the sensitive dog. Reading the dog correctly is equally necessary to the breaker or the shaper.
The shaping technique is just as useful in correcting bad habits as it is for establishing new good ones. To correct a fault the dog has acquired, such as busting birds, one might resort to the shock collar or the check cord. But better go to shaping first to unlearn (extinguish) the dash in part by withholding reward for doing it (no bird to chase, just a cage or even a scent so no mouth full of bird to reinforce dashing in).
Then reward for not dashing in, then reward for stopping. The shaping will take more ingenuity and patience and brains on the handler’s part, but the results will be more satisfying and longer lasting. For a dog that will surprise you with its apparent smarts and one that can handle unusual situations with absolute brilliance, go heavy on shaping and very light on breaking.
Updated 15 May 2005