The basis for the recall I teach my dogs is actually classical conditioning. When I call them, I want their bodies to be moving towards me before they have a chance to think about what it is that they are doing.
It sounds like this would take a lot of time, but it truly doesn't. It only takes five minutes to do 150 repetitions in the beginning. So if you do three sessions of five minutes each a day, you can get over 3000 repetitions in a week. In other words, in very little time you're talking about lots of reps.
Step One: choose a consistent stimulus to use as your recall command. For most dogs, a sound is best because they can hear it even when they're not looking at you.
For dogs that are deaf, there are two possibilities: a visual cue like a light flash or a touch cue like a remote controlled vibrating collar. The visual cue works well--but only if the dog sees it. The remote controlled vibrating collar is more difficult to acquire but it can be incredibly effective--even more effective, IME, than a shout or a whistle is for a dog with normal hearing. If you know that your dog is a unilateral deaf (deaf in one ear only), it may be more effective to use the vibrating collar. Being deaf in one ear leads to not being able to localize sounds very well and, especially at a distance, sounds seem to get bent somehow so that they don't sound the same.
Most handlers who have dogs with normal hearing will probably choose a sound. The great advantage to having a word is that it's always with you; the disadvantage is that it's difficult to keep the word sounding the same, which tends to blunt its effectiveness. A whistle or other mechanical noisemaker has to be carried with you but it always sounds the same. Each individual has to decide which way the balance tips: convenience of the word vs consistency of the whistle? Less effective word vs the inconvenience inherent in using a mechanical device? Each person differs so it has to be a personal decision.
I call this the Secret Stimulus or Secret Sound (SS). The reason it is a secret is that it must be something you use only when you are sure the dog will respond. It shouldn't be something you use in every day conversation or when you don't know how the dog will respond. You will be setting out to deliberately create a reflex in your dog's body; it will be much easier to get that reflex and keep it honed sharp if you are careful to limit the dog's exposure to the cue that stimulates the reflex
Step Two: figure out a list of twenty things that your dog wants. A hypothetical dog's list might look like this (not prioritized): *Cheerios *desiccated liver *strained baby food meat (if refrigerated it makes a nice stiff paste) *peanut butter *homemade crispy treats *Froot Loops *banana *apple *bits of steak *cream cheese *cheeseburger *squeaky rat toy *hedgehog *Fuzz-bee (fake fur disc) *fake fur snakie *tennis ball *canvas bumper *raw beef bone *a particular dog friend *a particular cat friend *chest scritches *butt scritches *picked up and hugged by handler
Some of these things are more portable than others--but they can all be used in at least one situation. To get in the most training, make most of the desired things as small as possible. My dogs will work for individual Cheerios or Rice Krispies (but hey, they think they're starving). An occasional jackpot might be a large hunk of food or several throws of the tennis ball or a prolonged game of tug with the fake fur snakie--but jackpots are, IMHO, of less effect in training than many tiny reinforcements. For example, imagine what you could get trained with one hot dog cut into sixty pieces versus the same hotdog cut in half.
With my dogs (and many other dogs) there doesn't seem to be much difference between one piece of food and five pieces of food given all at once. However, my dogs are much more impressed with those five pieces of food if I hand them out one at a time. The same thing may be true of humans as well.
If your dog is free fed, food may be less interesting to them. I strongly recommend feeding in meals for many reasons. If for some reason this just doesn't seem practical for your particular dog, you may have to use better value treats or rely on nonfood treats.
Each of the following steps takes about a week:
Step Three: go to the most boring room in the house. For most of us, this is the bathroom. For those of us who are female, this means that we probably have several training opportunities a day. <g> Close out any other animals but the one you are training.
Give the SS (Secret Stimulus/Sound), click and hand the dog a treat. You're in a boring room with very limited options and you have something your dog wants--of course your dog is going to respond correctly! Do this over and over. You can do 150 repetitions easily in five minutes. The sequence goes: SSclicktreatSSclicktreatSSclicktreatSSclicktreatSSclicktreat, etc. >
If you are using a word, say the word in a "calling" tone of voice-a bit louder than your normal conversational tone. You don't want to blast the dog through the wall with your voice but do make it louder than usual (this seems to make it closer to what it sounds like outside at a distance).
If you are starting with a clicker savvy dog, I suggest you click whenever the dog's body is in contact with yours.
Work on perfecting the chain. First the SS. Then the click. Then the treat. It should happen as quickly as possible but each part should be clearly separate. No part should over run the previous or following part. This is more difficult than it sounds!
Repeat this step as often as you can--ideally, at least three or more times per day for a week.
Step Four: go to the next to the most boring place in the house and repeat Step Three. Over the next week, go around to all the rooms in the house and repeat Step Three. Try to do at least three sessions of five minutes each day and try to go to every room in the house to do them. The dog does not have to leave you or do anything more than what they did in the bathroom.
Step Five: put the dog on a six foot lead and go to the most boring outside place you can find. Keep hold of the leash, so that the dog can never go more than six feet away from you. Ideally you are in a location so boring that the dog never does much more than give the briefest of glances away from you before you start and has their attention riveted to you after you start each session. Repeat Step Three (SS/click/treat). During this week, try to do at least one session of SS/click/treat in the house (off lead) and two sessions outside each day.
Step Six: put the dog on a six foot lead and go to a new place each day to practice Step Three (SS/click/treat). This may call for some patience if you have a dog who is highly distracted in new situations. You can make it easier by practicing a few other things first (like sits, downs, stays, heeling, etc) and then starting your SS/click/treat session in each new place.
Ideally, any single distractor should be far enough away that your dog merely glances at it and then turns their attention back to you. However, back here in the real world, it's not always possible to achieve this.
I do not recommend allowing the dog investigate the area on their own before you get their attention--that's backwards for effective training! More effective is the attitude "first we do what =I= want, then you get a chance to explore/sniff/greet other dogs, etc."
If your dog is very distracted in new places, you may want to improve the value of your treats. For extreme cases, I use gooshy treats that I can literally squish right into the dog's mouth.
If your dog is so distracted in new situations that you can barely get any attention at all, even by doing some preparatory work and with the most incredibly wonderful treats possible, then your dog needs some work on learning how to ignore distractions or on becoming de-sensitized to normal levels of distractions. De-sensitization in this case means taking the dog somewhere they are exposed to the distraction and just waiting for them to get tired of trying to get to the distraction. You may well have to take a lawn chair, a good book, sunscreen and water for both you and the dog. Just wait until the dog stops simply trying to get to the distraction or distractions and starts to pay attention to one or more other things. At this point, you can then use your clicker and treats to shape the dog into paying attention to you.
Continue to practice Step Three (SS/click/treat) inside the home as well. Start calling the dog from slightly greater distances - well, at least, initiate the session with the dog at a slightly greater distance! After you call the dog the first time, s/he'll probably be pretty glued to you for the next few minutes.
Put the dog on a six foot lead. Set up a distraction inside the house. Start at a distance far enough away from the distraction that the dog is aware of the distraction but not overwhelmed by it. Give the SS, wait until the dog has turned to look at you and is moving towards you, click, and treat.
One such example might be a bowl of treats on the floor. You've got the dog on the six-foot lead, so you can easily keep the dog away from the bowl. Put the bowl on the floor in one room and go as far away from that bowl as you can and still have it within sight. If your dog is EXTREMELY distracted by the thought of food on the floor, let the dog see you put a single boring treat in a bowl on the floor, then go into another room out of sight of the boring treat. Have much better treats!
Gradually work closer and closer to the distraction. When your dog is handling distractions well inside the house, start again outside with your dog on a six foot lead with the very same distractions.
If the dog gets so distracted by something that they aren't responding to you, you have several options. One option is to just wait it out. You have the dog on lead, so s/he isn't going anywhere. If the dog turns to you within a reasonable amount of time, click and treat. If the dog doesn't turn to you, use your lead to gently turn the dog away from the distraction--dogs aren't born knowing how to stop staring at a distraction and some dogs learn it fastest by being physically turned away from it. Another strategy to try is to gently tickle or tug on a part of the dog's body--ear, flank, tip of tail, tailfeathers, etc. Use a different touch each time! If the dog is insistent on staring at the distraction even if you're trying to turn the dog away, back away from the distraction.
Another strategy is to use the clicker to shape the dog into paying attention (stop working on recalls and work on attention instead). Start by clicking any part of the dog that is moving in your direction-even if it's just a hair on the dog's right ear being blown towards you! Gradually up your criteria for a move in your direction until the dog's whole body is turned towards you.
Step Seven: get or make a long line. It doesn't have to be particularly light but it should be strong. The length is dependent on the relative speeds of the dog and handler. Slow dog with fast handler: fifteen feet or so of long line. Fast dog with slow handler: 100 feet or more of long line! Plastic coated clothesline makes a nice long line but it can get heavy in lengths longer than 20 feet or so.
Parachute cord can be good--but get the military specs parachute cord, if possible (much stronger and better construction). You can buy cotton or nylon webbing--again, it gets heavy in the longer lengths. On the other hand, I know a certain Belgian Princess who used to tow 175 feet of cotton webbing and it didn't appreciably slow her down. <mock glare> Tie knots at intervals along the long line--maybe every 10 or 20 feet or so. Tie a bigger knot at the very end of the long line.
For dogs with possible neck problems or with real speed, a body harness that avoids pressure on the neck is strongly recommended. I use and recommend harnesses from White Pine Outfitters.
I strongly recommend the use of a Snap-back for any long line exercise. They are made out of bungee cord material with a ring at one end and a leash snap at the other end. You snap the Snap-back to the dog's collar and attach the leash to the ring on the Snap-back. I used to make my own shock absorbers out of bungee cords and duct tape--this is the same idea but lighter and better made. A shock absorber protects the dog's neck and also helps keep the long line from breaking (think about the difference between trying to pull a string apart by gradually increasing the tension on it and snapping the string by giving it a sudden sharp pull).
Take the dog outside to your familiar area and attach the long line (if you make your own long lines, you can also make a variety of lengths to suit different conditions). Drop the long line on the ground.
LEAVE THE LONG LINE ON THE GROUND!
For one thing, this protects your hands from rope burns. For another thing, most dogs keep half an eye on their handler, so you don't want to incorporate the sight of you with the long line in your hands in the overall recall picture for your dog (or the sight of you bending over to pick up the long line). The knots in the long line are there so that if you have to stop the dog, you can step on the long line and the knot(s) will catch under your shoe.
Let the dog get minorly distracted, give the SS, wait until the dog is approaching you, click and treat. By minorly distracted, I mean at the level where the dog is glancing away but not focusing intently on any one thing. Repeat! Repeat! Repeat! Did I mention you do lots of repetitions? <g>
Over the next few weeks, move to the outside locations you've practiced in before, from most boring to less boring and repeat. The more locations you can repeat this in, the better the dog's recall will be. Start with locations that are familiar to the dog. After you have run through that list of locations, move on to new locations. Again, try to start with boring locations. Life being the way it is, you may miscalculate the distraction potential of a new location or something may happen.
If you give the SS and the dog ignores it, step on the long line. If the dog is moving fast, try to step on the long line in such a way that you gradually increase the tension on it or else use a shock absorber. Walk down the long line while trying to pretend you're doing nothing of the sort. As far as your dog is concerned, they've mysteriously become unable to move away. When you get close enough to the dog that you are certain your dog will respond, give the SS, wait until the dog is moving towards you, click and treat. If you can't possibly get close enough to the dog for them to pay attention to you, go back and build attention (see Step Six).
You can start leaving out the click. If I have a clicker on me, I use it--it's like getting two treats out of every treat. First the dog hears the click, which gives them a little jolt of pleasure, then they actually get the treat, which gives them a little jolt of pleasure. Two little jolts of pleasure are better than one little jolt of pleasure. If I don't have the clicker, I don't worry about it.
Step Eight: start deliberately setting up the situations that your particular dog finds most distracting. For example, one common strong distraction for most dogs is seeing other dogs, particularly other dogs that are playing. Start out at a distance away from the two dogs where your dog recognizes the distraction but is not strongly affected by it. The distance at which your dog glances at the distraction but does not lock onto it. Over many reps decrease your distance from the distraction without going over the threshold of a casual glance. If the dog locks onto looking at the distractor, move further away, change your reinforcer or re-evaluate what is going on.
As you get really close to the other dogs, the ideal set up is with someone who has two dogs that are good with other dogs and have perfect recalls or drops. That way, if your dog doesn't pass or leave the other dogs at some point, the other handler can call the dogs or have them drop (which almost instantly makes them boring--plus, it's hilarious to watch the trainee's reaction to this "did I say something wrong? Is it my breath?"). Less than ideal but workable is the other dog on a leash or behind a secure fence.
Start out by doing a few recalls with your dog on a six-foot lead far away from the other dog(s). The other dog(s) should be calm at this point--maybe doing sit stays or just hanging out calmly with their handler. Gradually work closer to the other dog(s) until your dog can do the recall less than three feet away from the other dog(s). Then put your dog on the long line and move far away again. Gradually move in closer and closer.
Once your dog can do the recall on a long line with a calm dog or dogs, then add in a little movement of the other dog(s). Move back far away and if your dog is particularly attracted to other dogs, go back to the six foot lead. Gradually work closer, then switch to the long line and move far away again.
Gradually up the amount of motion on the part of the other dogs.
Set up as many of your particular dog's distractions as you can. Kids, kids playing, kids on bikes and/or skateboards are both easy to find in most places and easy to set up (most kids love to help you train your dog). Use as many different locations as you can.
Step Nine: repeat Steps Six, Seven and Eight off lead.
If something goes wrong after Step Nine:
Now, by this time, your dog has probably done 22,204 or more recalls and been reinforced over and over. If you have done this right (have figured out what your dog wants as a reinforcer and have figured out how to set up distractions and use them so that your dog can handle them), you will have a dog that automatically turns and starts to head for you when you give the SS. For some dogs, this is all you have to do to have a reliable recall. Lukie the Boxer Puppy (anyone who was on the list a year ago will remember the Boxer puppy I was allowed to foster for a couple months) is now a big, gorgeous, intact male Boxer--who has yet to refuse a recall.
Howevah! There are some dogs who will turn and then go "naaaaaah, I don't think so" and deliberately go back to what they are doing. Dogs do have the capacity to weigh their options and make the decision I don't want them to make.
I am not a clicker fanatic. I am a fanatic about good recalls.
My first use of an aversive is to have the dog on a long line, step on the long line to restrict the dog's ability to move away, then I walk the dog down, take the dog's buckle collar in my hand and make them come with me as I walk backwards away from whatever was distracting them. I walk backwards because that's the side of my body they usually see when they're doing a recall (not the side of my body). No verbal scolding, no scowling! Do this with as little emotional impact as possible. When I've moved far enough away so that the dog has functioning brain cells again, I drop the collar while continuing to move backwards. If I've judged my distance correctly, the dog continues to follow me. I click and treat. Then we try that distraction again, but probably at a greater distance or a less distracting version (for instance, instead of all six dogs playing, have them all sitting or standing).
If the dog is still refusing the recall, then I escalate my aversive a bit--usually, to taking two big handfuls of ruff as I move backwards. If the dog is able to remain oblivious to a pull on their collar, they may be what some trainers refer to as "neck dead." These are dogs that either by nature or conditioning no longer respond to touches or collar sensations on their neck. Such dogs are highly unlikely to be unduly upset by the handler taking two big handfuls of ruff. Many of these dogs probably play with other dogs by being dragged around by the heck or even flung into the air by the neck. As a human being, I cringe in sympathy pain when I see this but many dogs not only tolerate this, they enjoy it and actively solicit it. This is not the sort of dog that is going to find being led by two big handfuls of ruff to be painful.
Again, no verbal scolding, no scowling! Be as neutral as possible. Pretend you're an animated fencepost. You are trying to get attention and remove choices but not trying to scare the dog or anything like that.
If I'm getting more refusals, I stop and think about the situation. What part doesn't the dog understand? I have to fix that part.
Some dogs do get "long line wise." That is, they never refuse a recall while on the long line but when the long line is gone, they are unreliable even if the handler started with close recalls off lead in a familiar (and safely fenced) area.
Some dogs become "fence wise." They never refuse a recall when in a securely fenced area under a certain size (usually an acre or so) but in a larger area do not have a reliable recall. Generally the dog knows quite accurately how fast the handler can move and will be reliable in fenced areas small enough that the handler can catch them quickly but in a larger area does not have a reliable recall.
IF the dog never ever refuses a recall while on the long line or never ever refuses a recall in a securely fenced area, no matter what the distraction (and I am pretty good at devising distractions for this test!), then I would move to a shock collar. A good test for the dog would be to pass three of the most highly distracting situations the handler can devise.
So far, in approximately nine and a half years of teaching the Recall Redux, I have never run across a case where a dog that has gone through the entire training program conducted properly has needed a shock collar to gain a reliable response to a recall. However, dogs are living organisms and I do not feel I have enough data gathered yet to say that this is true of all dogs.
Yes, a shock collar is painful and aversive. However, I've been bumped by cars on three different occasions and I've tried out shock collars set at the highest level on myself and I can testify that being hit by car hurts much worse. A shock from a shock collar adjusted correctly carries no risk of permanent physical harm while being hit by car carries a very high risk of permanent physical injuries or death. The leading cause of broken bones in dogs in the USA is being hit by car. Almost every general practice small or mixed practice veterinarian can cite cases of dogs that have been hit by car more than once, so being hit once and surviving is no guarantee that it will never happen to that dog again.
While I certainly believe that management in the form of leashes and fences is helpful and has a place in training dogs, all my life experience tells me that management alone always fails. I have had dogs jump six foot solid wood fences without touching them and known dogs that scrambled over eight foot chainlink. I have had dogs break leashes, break collars, chopper leashes, had leash snaps break, dogs slip collars and the most common failure of all--I've fumbled and dropped the leash at least ten thousand times. Rely on management alone long enough and management will always fail. One hopes that the failure is not catastrophic but there's no guarantees.
To me, the choice is clear. If I were sure that I had done the entire Recall Redux, if I were sure that I had done so at the very best of my training skills and my dog still did not have a reliable response to a recall, then I personally would feel that I had reached the point where a catastrophic failure of management was likely. I'd rather do something painful and aversive to the dog than have my dog killed by car. However, each person must make their own decision that fits their own ethics and lifestyle that is consonant with their own assessment of the relative risk levels.
I offer the following brief outline on one strategy for using a shock collar in the belief that having as much information available as possible is generally a good thing. Handlers who are unfamiliar with shock collars should seriously consider finding a trainer experienced with the use of a shock collar for instruction and supervision.
In the following tests to assess whether there is little likelihood of further training via classical conditioning and positive reinforcement will improve reliability, a pass is where the dog whips around on cue without any perceptible hesitation and moves at a high rate of speed directly to the handler. Any hesitation or deviation from the straightest possible course to the handler at less than high speed is an indicator that the dog needs more work in the associated area(s).
Unless the dog has a highly specific and predictable distraction, try to choose tests that cover as wide a range of situations as possible. Not just food, not just possible prey, not just toys, not just other dogs--try to test all of them!
Some examples of distractions to test might be:
- whatever that particular dog has found most distracting in the past. Call the dog when the dog is actively involved with that distraction.
- a chicken in a simple Figure-8 loop body harness which can be homemade out of cord tied to a stake and provided with a small amount of grain so that it is moving around. Call the dog when the dog is less than a foot away from the chicken with the handler stationed at least fifty feet away. Do not use a chicken if it would break your heart for the dog to kill it! The dog can be on a long line but it may not be enough to prevent the dog from killing the chicken.
- a large chunk of raw liver wrapped in wire and attached to a screw in post in the ground such that the dog can nibble and chew on it but cannot remove the bulk of the liver at one time. Wait until the dog has been chewing on the liver for a minute or two, then call the dog with the handler stationed at least fifty feet away.
- a group of kids playing soccer and actively including the dog in the game. This test should be conducted in a safely fenced area that the dog has had ample time to explore first and check out the fencing. It cannot be conducted with the dog on a long line because of the high risk of getting someone tangled up and rope-burned. Let the dog play with the kids for at least five minutes, then call the dog with the handler at least fifty feet away while the kids continue to run around and use body language that invites the dog to continue playing.
- a group of friendly dogs playing. This test should be conducted in a safely fenced area that the dog has had ample time to explore first and check out the fencing. The test cannot be conducted with the dog on a long line because of the high risk of one or more dogs getting tangled up in the line. The dogs should be allowed to play for at least five minutes, then call the dog being tested with the handler at least fifty feet away. The other dogs are allowed to continue playing.
- a cat in a wire crate. Choose a cat that is placid and unlikely to be terribly upset about having a dog run up to it. Let the dog see the cat from 25 to 50 feet away while the handler deliberately pulls the dog backwards. Then the handler lets the dog go and stands still. When the dog is within a foot of the crate, the handler calls the dog.
- a person that the dog knows well but does not live with the dog sits on the ground with a big container of treats and gives the dog rapid-fire treats as well as petting the dog. The handler is at least fifty feet away. When the dog has been interacting with the helper for at least three minutes, the handler calls the dog.
- dog is wandering around with a well known person who does not live with the dog present but not interacting with the dog. This test must be conducted in a securely fenced area that the dog has had ample chance to explore and check the fencing due to the risk of tangling the helper in a long line. The handler and helper deliberately stay far apart, a minimum of 25 feet, while wandering around casually until the dog is ignoring the helper or at least is not sticking close to the helper.. When the helper has managed to move into a position between the dog and handler, the handler calls the dog. As the dog approaches the helper, the helper pulls out treats to shower on the ground in front of the dog, the dog's favorite toys to toss around and uses verbal encouragement to try to stop the dog from reaching the handler.
To keep the dog from becoming collar wise, I spend a couple weeks fiddling with putting both the dummy collar and the working collar on and off the dog. I put it on several times a day and leave it on for a couple hours at a time. Every time the dog goes outside, I put the collar on (even if I have to take it off first in order to do so). Basically, I make such a production out of putting the collar on and off that the dog starts to tune it out.
There are two ways to use a shock collar: one is as a positive punisher for any behavior that isn't a recall (after the SS, of course!). The other way is as a negative reinforcer (the collar is set fairly low, handler gives SS, then starts shocking dog, stops shocking the dog when the dog has arrived). I use the collar as a positive punisher.
To be effective as a positive punisher, I set the level of shock at a level high enough (in my best guesstimation) to be unpleasant to the dog. I want the dog to startle a bit and even yelp when they get shocked. At any point where the dog deviates from the straightest possible path to the handler, the dog is shocked. No, this is not pleasant. After each shock, I call the dog again (because many dogs tend to panic when something mysteriously reaches out and stings them).
Done properly - first training the dog to do the recall under many different conditions until the dog never ever refuses a recall when they are attached to a long line, and with the level of shock set high enough for the dog to actively want to avoid it - I have found that the shock collar is incredibly effective. It doesn't take many shocks (far less than ten) before the dog is equally reliable off line as they are on the long line. This was before I tried out classically conditioning the recall. I have not needed to use a shock collar since 1992, when I started classically conditioning recalls.
After the dog has gotten to the point that they never refuse the recall, I start to phase out the shock collar by using the dummy collar again. Sometimes the dog has on the working collar, sometimes the dummy collar. I keep messing with and switching the collars several times a day, separate from any training situation.
And that's how I get a reliable recall. I would say that 90% of all the dogs I have personally taught a recall to or supervised the teaching of the recall either never had an aversive or only had the led-back-by-the collar aversive fewer than five times. The vast majority of dogs belonging to clients who came to me more than nine and a half years ago specifically in order to use my shock collar never used the shock collar--when the owner goes back and re-taught the recall, the problem disappeared in the vast majority of cases.
It's been so long since I had to use my shock collar (and a couple moves!) that I no longer even know where my collar is.
There may be a few dogs who do not become reliable--and that's where the skilled use of aversives can save their lives. I believe that at least part of what may be happening with such dogs is instinctive drift - they are being asked to stop doing something instinctual (coursing prey or mating rituals) and that instinct is stronger than the handler's ability to classically condition a conflicting response.