From my Front & Finish Column in the 90’s.
As promised, this month we are going to start building a two-part “down-stay” from start to finish, clicker style. For those of you who are instructors, I am going to digress a little and actually describe the initial shaping. Teaching a “down-stay” will obviously go smoother if the dog already knows the relationship between the click and the treat – if you are teaching classes or doing private appointments, you can use this exercise to make the click = treat association while shaping the behavior. This month we will take the first stages of shaping the behavior and varying the reinforcement, then finish up, next month with creating reliability and integrating the behavior into the dog’s repertoire.
NOTE: With most clicker trained behaviors, 99% of the job is making sure the dog really understands the task. As much as possible, pushing, shoving or tugging should be avoided. It isn’t that pushing a dog is going to traumatize the little beast, it’s that using your hands is an extra step that will take additional time to eliminate, later.( If the only practical way to get the behavior is to gently “model” the dog into position, do it – then click and treat. Over a series of repetitions, soften your touch until the dog no longer needs the tactile hint. )
First shaping: Trainer in a seated or kneeling position. Dog in a treat-seeking, preferably seated position, facing the trainer.
Take a favored food treat and move it up to the dog’s nose. Move the treat straight down to the ground, in between the dog’s forepaws. I mean straight down, not angled out toward you. Imagine letting go of the treat – if it fell to the ground, where would it land? That’s the place you want your hand to be at the end of the first movement. Once the dog’s nose catches up with your hand, click and treat.
The next step is to teach the dog to be comfortable following your hand to the ground. As your hand hits the ground, pause for a second and allow the dog to catch up with your hand. You may even let the dog lick or nibble at the treat. After the dog will consistently follow your hand to the ground, follow the short pause by slowly moving your hand along the ground, toward you. If you viewed this from the side, you would be tracing the shape of an “L”. (The down stroke ß is followed by the ground stoke. Þ ) If you do this correctly, the dog should be fully down within ten repetitions or less. If the dog stands up each time, go back and reinforce the “down stroke” of the “L” a few more times. If it takes a few more repetitions to get a consistent down, don’t worry. There are many dogs who will not have read this article and won’t quite know what you are driving at. Caution: If you attempt this with a Mini-Dachshund, you should probably examine your motives.
The “Bait and Switch”:
Once the dog will follow your hand to the ground, try to get ten repetitions in a relatively short period of time. Your goal is to build some momentum so that you can pull the old “Bait and Switch” on the dog. That means that you are “baiting” a pattern that the dog will come to anticipate. By utilizing the dog’s natural ability to anticipate, you can later “switch” the rules and trick the dog into making a mental leap to the next level of expertise. Here’s how it looks.
After your tenth repetition, touch the treat to the dog’s nose and quickly pull it back to your chest. You have just done ten in a row with your hand moving to the ground and the dog going to the ground. The momentum is all on the side of the dog “going down.” If you do this tease correctly, the dog realizes two things – 1) You really do have a treat and 2) You didn’t move your hand the way you did a moment ago. If you have created a good pattern, the dog will “mistakenly” attempt to do some part of the “down” immediately after you tease his nose with the treat. If the dog tries to stand up to investigate or grab the treat from you, say wrong and tease the dog again.
As a general rule, when you are trying to tease the dog into the behavior, try no more than three teases, of no more than 20 seconds in between each one. If the dog hasn’t offered any part of the behavior after a minute of teasing, the behavior isn’t strong enough, yet. Go back and do ten more repetitions of luring the dog to the down position before you try the teasing again.
Whether it takes one or two extra sets of luring, you will eventually be able to “tease” the dog and see some part of the behavior occur. This may simply be a head dip, or a slight forward movement of one paw. Click and treat! Take any tiny piece of the behavior the dog is willing to give you. You are trying to help the dog to re-create the behavior without the assistance of your hand movement. If you continue with this process, on one of the repetitions, the dog will drop – which may cause you to smile. Once you have succeeded in your first tease-down, do at least 10 more successful repetitions.
BUILDING DURATION AND STAMINA:
After you have your first tease-down, add these additional criteria in sequence, over a series of sessions. (Remember, if the dog breaks, say “wrong” in a neutral tone of voice and start another repetition.)
- The dog will go down when you are kneeling.
- The dog will go down when you are standing.
- The dog will go down and stay down while you turn your back for 10 seconds. (This is a great time to have a helper who can operate the clicker and say “wrong” for you. If you always cock your head over your shoulder to make sure the dog is immobile you are merely postponing the inevitable. Eventually you have to teach the dog that “stay” means “even if I am looking/walking the other way. The best solution is to teach this distinction from the very beginning. )
- The dog is not dependent on being right at your feet in order to go down.
- Your dog will hold a down while you walk in a circle around him/her. This is the same thing we all learned in basic training 101. The only difference is that you aren’t getting on the dog’s case if he/she breaks. You are clicking for staying and “wronging” for movement.
Adding a Cue – finally:
Now that you have a dog who will drop at the drop of a hat, it’s time to attach a cue. If you plan on working this dog in obedience, or any situation that demands great precision, attaching the performance cue should be one of the last things you do. In order to give you a way to trigger the cue while you are building all that wonderful precision, attach some other word to the behavior for now – a word you are unlikely to use in performance. The “working” cue can be anything you like, so don’t feel limited. The point of using a working cue is to make sure that the performance cue is never associated with the inevitable sloppy versions of the behavior that are a necessary part of training and learning.
When you are confident that the behavior will occur on a predictable schedule, start saying the word, just before the dog does the behavior, then click and treat. Since you have been getting the behavior consistently, your dog should be offering it on a cycle. After each click and treat, there should be a brief pause and then the animal offers the behavior again. The time to inject the new cue is after the presentation of the treat (for the last repetition) and BEFORE the dog initiates the next repetition. If you do this 20-50 times, the dog will automatically start to make a connection between the cue and the behavior. If you haven’t done this before, this may sound a little odd, but dig up my column on “cues” from a few months ago.
Now that you have a cue that can “sort of” trigger the behavior, it’s time to start flexing your dog’s mental muscles. Pick a time when your dog is standing at a short distance and facing away from you. Give your new cue and see what happens. If the dog offers any part of the behavior, click and treat. If the dog makes any move toward you, say “wrong”.
Once you have succeeded in beating the distance barrier, start practicing your dog’s new knowledge everywhere you can. If your dog is standing in the kitchen facing away from you, say “down” in an normal tone of voice and see what happens. While it is best to be prepared and carry a clicker with you for this phase, if you don’t have a clicker, you can always “Good”. If you don’t have a treat, use praise and affection.
Varying the Reinforcement:
After a behavior is recognizable, it’s time to start varying the reinforcement. To refresh your memory, go back and look at the column that dealt with getting “R.E.A.L.” When you start this process, remember that varying the reinforcement will cause variable behavior. The carefully crafted criteria that you have spent a week teaching will get a little flaky for awhile. That is normal and desirable. If you don’t see any “wiggle” in your dog’s behavior when you vary the reinforcement, you are doing something wrong.The purpose for varying the reinforcement at this stage is two-fold. First, you are trying to get the dog to experiment and improve his performance. Second, you want to make the behavior stronger and less dependent on constant reinforcement.
For instance, in order to make your current (marginally acceptable) down start speeding up, you have to make the reinforcements less predictable. By changing the type, rate and quantity of reinforcement, the dog will start offering variations on a theme. When you see an aspect of the behavior that exceptional, give the dog a big jackpot and pour on the baby talk and affection. While you might assume that this jackpot is meant to “That version of the behavior is worth far more than other versions,” there are two other reasons that are more important. First, a large, unexpected jackpot will often cause your dog’s behavior to become more variable. That’s what this exercise is all about. Unless you can get your dog to change its behavior, you will be stuck with mediocrity. The other purpose for this jackpot is to make the transition from consistent reinforcement without causing the dog to quit trying. The jackpot is intended to keep the dog working through the dry spells that will result as you raise your standards. i.e. Now that your expectations are higher, you will naturally reinforce less often. Unless you increase the average reinforcement, the dog may quit working before you have communicated the new criteria.
One of the easiest ways to vary the reinforcement is to make the length of the stay unpredictable. There is a tendency for most people to steadily increase the amount of time the dog must stay. For instance, once the dog will do 30 seconds, the trainer tries for 45 seconds, then a minute. If you escalate in an ever-increasing fashion, your dog will soon think of the process as drudgery. A better approach is to ping-pong the time so that some repetitions are short, while others are longer. For example, here are ten repetitions that gradually extend the stay from about 10 seconds to about 40 seconds. 1) five seconds 2) 2 seconds 3) 8 seconds 4) 30 seconds 5) 22 seconds 6) six seconds 7) 33 seconds 8) 21 seconds 9) 9 seconds 10) 40 seconds. You can see that the amount of time is unpredictable and that sometimes a long period is followed by an “easy” stay of only a few seconds.
NOTE: Remember, the click ends the behavior. During this exercise you may be tempted to click in the middle of a stay in order to keep the dog “remaining.” This is a serious error. If you start misusing the clicker to mean “end of behavior” and “keep going” you will only confuse the dog and create hidden weak spot in your long downs and sits – hidden until you need the behavior most. If you absolutely must have a signal that means “keep going”, simply say “Keep Going.” (duh) If you use a “keep going” signal you must make sure it is eliminated long before you get to performance levels. I do not recommend using “keep going” signals at all. The bottom line is that the behavior is over when you say it is over – period.
After you start varying the reinforcement, you should start seeing variations in the dog’s behavior that will allow you to capture faster, prompter, crisper downs. Now go back an re-introduce the other important criteria such as “down from a distance”. Gradually polish the behavior until you have a reasonably quick down, from a distance. It may not be exactly straight, yet, but it should be pretty solid. Start extending the ratio of behaviors to reinforcement so that you can get at least four clean “downs” between clicks and treats. For the middle repetitions that aren’t reinforced with actual treats, continue to use the click to mark correct behavior, but use affection and praise as the reinforcers. If any of the behaviors are substandard, say “wrong” and end the sequence. If you are getting lots of “wrongs”, drop your standards, reinforce more often and re-build the consistency before you go back to varying the reinforcement. Next month we will look how to take this behavior to a performance level, how to proof and maintain it.
Note: I am gradually collecting and publishing articles from my Front & Finish Column. They will be added and collated over time.