Anticipation - A powerful training tool

Dogs anticipate. Motiviated, driven dogs anticipate more then their less driven counterparts. It is a natural part of the learning process for the dog and most competitive trainers should not only expect it but make it part of their training repertoire to sharpen up their dogs performance.

Almost all of us use anticipation when we use classical conditioning to teach our dogs verbal commands. It goes something like this:
- Lure the behaviour you want to teach (e.g. move hand with treat from dogs nose, down to the ground between its paws to teach it a platz)
- Fade the lure away so that a hand signal is firmly established for the behaviour.
- Now you are ready to teach a verbal command. Say the command "platz" first, THEN, do the hand signal to get the dog into the right position.

Some repetitions of this and your dog will platz on command before you do the hand signal, because it will anticipate the hand signal that it has learned for he behavior.

This is simple classical conditioning as we have all learned from Pavlov. When Pavlov rang the bell after the food was served to his dog, or while the food was served to the dog, the bell had no significance. It is only when the bell rang first and THEN the food was served that the dog associated the bell with food.

The odd thing is that while we all tend to know how to apply anticipation in this situation - to teach our dogs verbal commands - soon after, we forget that this is a powerful tool that can and should be used in training. Instead it becomes the thing we try to teach our dogs not to do by correcting them for it:
- Dog anticipates a "HIER" from a platz position, we start correcting it while putting it back into position.
- Dog forges ahead to walk into escape position after the helper has stepped out of the blind, bang, bang bang. It is incredible the number of dogs that forge to the point of crowding the handler walking from the call out position to the escape position. The only thing that seems to go through the handlers head is "I said foos.... the dog is forging.... correct, correct correct". It never works but the dog gets corrected session after session after session and looks the same at trial time.
- Dog automatically platzes when we have stopped at the escape position instead of sitting and waiting for the helper's command to platz, BAM. Dog gets yanked off the sit with a pinch.

Don't get me wrong here. I am not against correcting dogs. In our sport, with high drive dogs, positive punishment has its place. However when you do the same thing ten times and the dogs behaviour does not change, it is counter productive to keep doing it over and over. Ultimately the dog desensitizes to positive punishment and then you start looking for ways to up the ante with the punishment. This is a never ending cycle and nearly always causes conflict in the dogs mind. Conflict is not helpful as part of the training process. When corrections are used, we want them to be fair, crisp, precise and most importantly we want them to modify the dogs behaviour in the way we wish.

So what does blurb on positive punishment have to do with anticipation? A lot. For starters, we tend to correct quite a bit for problems that are simply caused by the dog anticipating the next step of the exercise because we have done it again and again and ... again. Like I said, dogs anticipate.

A far more effective method is to:
1) Anticipate that your dog will anticipate the next phase of each exercise if you put it together.
2) Modify your training to either prevent your dog from anticipating, or better yet, use the fact that it will anticipate during your training.

I will walk through how I do this to produce a correct foos from call out position to escape position.

To start, I nearly never actually practise escapes asking the helper to step out of the blind, foosing to the escape position and platzing the dog. Its counter productive because I KNOW that if I do this a few times in a row, my dog will start anticipating each piece and then I will be fighting with my dog to get it to do what I want it to do. These are all separate exercise that I train separately and practice separately the vast vast majority of the time. One out of ten times I will put two or more parts of the exercise together.

Asking the helper to step out of the blind, foosing to the escape position, platzing my dog there, are three different exercises for my dog. When I ask the helper to step out of the blind, the behaviour I want from my dog is for her to stay calmly in control in the basic position while the helper gets ready for the escape. To help my dog look good, I will frequently start the exercise with my helper in the blind ( no revier). My dog in basic position beside me. I ask the helper "Step out of the blind". He walks and gets ready for the escape. My dog is still in basic position. Now after the helper has set himself up for the escape, he comes back a couple seconds later and gives my dog a bit on my authorization (I use "OK as a release world). My dog is thus rewarded for staying calmly in basic position. I can do this 50% of the time when I ask the helper to step out. My dog stops guessing what is going to happen and calmly stays where she is supposed to be ... in control... because occasionally, she gets rewarded there.

Similarly, with my helper ready for the escape, I will foos my dog over and the vast vast majority of the time, I will foos in a direction AWAY from the helper before I turn around with my dog and take her to the escape position. In fact the vast majority of the time, when I say foos, I make a left about turn or a hard left turn (If my helper is to the right). I want to err on the side of my dog anticipating a turn away from the helper because in trial this will help my dog to not forge when I start walking towards the helper. Sometimes I will ask the helper to come back while we are facing away and giver her a bite in basic position. Sometimes I heel to the escape spot and with my dog still in basic position, I ask the helper to come and give her a bite there. This teaches my dog not to anticipate the platz because she gets rewarded in the basic position for performing the behaviour she is supposed to, rather than the behaviour which she things she needs to perform next.
I like to do these bites on my authorization so that the dog understands that me the handler has the ability to give the dog what she desires - the sleeve is the reward. I am not simply an entity that is in the way of the dog getting the reward.

By now you are probably wondering "does this guy ever practise an escape?" Sure I do. However, I rarely do it after practising the setup that I am referring to above. When I want to practise an escape, I go a spot on the field and platz my dog, then I ask the helper to get into position. Now I walk away and the helper takes off for the escape. The escape itself is a different exercise to the dog from the three steps in the setup.

The key to making the performance look good is to break exercises into tiny little pieces and practicing them separately during training and occasionally putting different pieces together. It takes a certain amount of patience to reward all of these little things but if you are after the minutiae as I am, to get a picture that looks correct, then you must train and reward the little things and put them together from time to time and then all together only once your dog is performing each piece perfectly. Even after your dog is a SCH3, you need to keep practising and rewarding the little pieces if you want to keep your dog from looking poor because of anticipation which you did not anticipate.

I also use anticipation extensively during obedience. A simple example is always doing a group before a sit in motion. I do a group around objects in my yard before a sit in motion. I never do a sit in motion without it. This allows my dog to anticipate the exercise that is going to follow and makes it more likely that her sit will be fast and correct if she knows it is coming up. If you use anticipation in this way, then one day your dog will sit without you asking for it during the build up for the sit in motion. ANTICIPATE THIS! Your dog is simply showing you that it is pretty darn smart and responds to classical conditioning. You knew this right? After all classical conditioning is what you used to teach verbal commands. Too many handlers fall apart when something like this happens. The appropriate response when your dog anticipates and does something like this is to simply mark it with a no reward marker (I am assuming you are using operant conditioning) - such as "WRONG", go back and do the exercise over. To prevent this type of anticipation, vary the number of steps you take in training during the build up of the out of motion exercises. Sometimes do 8, sometimes 6, sometimes 20. Another example is that I never do a stand out of motion without having my dog come into foos postion from a front. Since I reward my dog very frequently in the platz position in the out of motion exercise and train the recall separately to prevent anticipation, my dog is frequently not in a platz away from me when I am ready to do a stand in motion. So I simply get my dog into a front. I have taught my dog a front command. I always start my stand out of motion from at least this position. I say "foos", my dog gets into position and we do a stand in motion.

A very powerful use of anticipation in obedience is to get crisp out of motion exercises without the forward creeping that sometimes occurs after the command. You use, after doing out of motion exercises over and over and over ..... and over, your dog realizes that after you give the command, you keep moving forward, and forward and forward. Much of the time you have done this in the past, you have not had eyes behind your head and no you have not had a mirror around or a spotter because you are doing this on your own on a weekday morning. It is impossible to correct your dog once your dog is already in the down position after having crept into the down, while you have already walked ten steps ahead. Well, you can correct him all you want, however, it is not bound to be productive. Corrections that are not productive simply cause conflict. The method that I use for this is simple, it has two parts:

1) For starters I teach my dog that when when I give it a command out of motion, it should anticipate backward, not forward momentum on my part. This means, I say "PLATZ", then one second later, I stop walking forward and I walk backward in the direction of the dog and pass the dog walking back ward. Sometimes I reward the dog when I get to it. Other times, I walk past it and come back and reward while it is still in PLATZ position. I prefer to walk right up to the dog and deliver the reward while it is still in position. I do this 7-8 out of 10 times when I do an out of motion exercise. If I teach my dog to anticipate backward motion on my part right after the command, rather than forward motion, the command actually helps the dog stop dead and go into the appropriate position. I still have the opportunity to work on things like my dog creeping after being in position (that is a different problem from the one I have just described). I can do that separately by simply placing my dog in the appropriate position (no out of motion) and then walk away from my dog as I would if I was doing an out of motion exercise. If I do this, then I do this with a spotter or a mirror in front of me. That way the microsecond the dog creeps, I can mark it with my no reward marker, go back to my dog and redo the exercise. I don't get mad, I just pick up the dog and set it up again.

2) Secondly, most of the time I want to practise an out of motion, with me continuing to move, without a spotter or a mirror, I do it with my dog walking with me while I walk backward. This way I can see exactly what my dog does and if it creeps before going into position then I mark it and restart the exercise. Rewarding the precise moment your dog stops dead and does the exercise well also helps and for this you need to be looking at it. I have found that walking backward or looking into a mirror ahead of you is the only way.

Some trainers use a leash around the dogs belly to pull up on the dog's belly to teach the dog a fast stand on the spot. This also uses the principle of anticipation. You say the command before you pull up on the lead attached to the harness. The dog learns to anticipate the jolt up of the harness and stops on the spot. This exercise too is useless unless you say the command BEFORE you yank on the belly strap. I am not adverse to this method but I don't like messing around with additional equipment. However, I have to point out that 9 out of 10 people that I see using this say the command at the same time the yank the belly strap and Pavlov taught us a long time ago that this is entirely useless. You need to create a classically conditioned response to the command if you are going to use an aid to help your dog stop on the spot - regardless whether that aid is a belly strap or you walking backwards after you say the command. If you use the aid at the same time as saying the command then the aid is useless and you might as well not have it.

There are a vast number of places that you can use the fact that your dog will learn to anticipate exercises that are put together in Schutzhund. Each part of the exercise can be used as a secondary reinforcer by simply rewarding the dog with the primary reinforcer immediately after the secondary reinforcer you wish to establish.

For example, my GSD Meika, is not a particularly driven dog and prone to distractions especially as I approached and went around the group. To counter this tendency, I established the group, or approaching the group, as a secondary reinforcer. I simply rewarded a few feet a way from the group, as I approached the group. After doing this 10 times, she was convinced that once I got within a few feet of the group a reward was imminent and thus the group served as a secondary reinforcer. Then I switched to a variable reinforcement regimen so that at least 50% of the time, she continued getting rewarded as we approached the group. Other times she was rewarded in the group during the figure 8 and so on. The result was a fairly low drive distractable dog having fantastic focus in the group. As I approached the group in trial, her intensity went up not down. Her focus on me was superb - particularly in the group.

Similarly, to prevent a dog looking around while its handler walks up during the hold and bark in the blind or guarding phase after an out, I like to make the handler walking up a secondary reinforcer that predicts the reward (bite). I used this extensively with my dog Farrah who has about as intense a silent guard as you can expect any dog to have. To get this, I simply have the helper reward the dog with a bite as I a approach the dog before I get to the dog. Once this is established, I have the helper give the dog a bit after I get into basic position. My dog learns to anticipate that me walking up to her is a reliable predictor of a bite, rather than a predictor that I will try to pick her up and foos her away which is what 9 out of 10 handlers do over and over and over during training.

To a dog doing a guard after the out, and then heeling away with the handler on foos once the handler arrives, are two separate exercises. You can and should teach them separately and put them together sparingly. More with some dogs and less with others.

Using these techniques, a dog that would otherwise look week in the guard by looking around for his helper, looks strong and intense in the guard. Of course this dosen't mean you only give the dog a bite when the handler is beside it. Do that to often and the dog will learn to predict that he only gets a bite when the handler is there and its guarding will be lackluster otherwise. The thing is, that during training, dogs naturally do get to rebite after an out and so this enforces good guarding behaviour when the handler is not there. On the other hand, when the handler approaches, most handlers simply want to put all the steps together and the dog learns that the handler approaching is a pretty reliable indicator that it will be picked up and heeled away from what it desires (the sleeve). 9 out of 10 times this is why dogs start looking around for the handler during the guarding phase. Nervier, less independent dogs do so more than others.

You should anticipate that when you use secondary reinforcers this way, a dog will occasionally take a dirty bite out of frustration. This is simply a apart of the learning process. The helper should be ready for this and mark the behaviour. This is one example of a case where I *may* use positive punishment in the form of a stern "NO". Most dogs will figure out that this behaviour is incorrect. An alternative of course is for the helper to disengage in addition to a no reward marker (drop the sleeve and look away) and have the handler pick up the dog and repeat the exercise (a.la. Ivan Balabanov's the "Ultimate Punishment").

I prefer not to correct for anticipation. Positive punishment extinguishes behaviours. I like a dog that offers a lot of behaviours and feels comfortable doing it. I find that correcting anticipation usually causes conflict and stress. I'm not saying that I would never correct for it, just that I prefer not to. I find that something like a no reward marker that clarifies that the behaviour was incorrect, works much better. This method of course works for motivated and driven dogs that have a strong desire for whatever reward we possess. In the protection phase when the dogs drive and desire for the sleeve are high, I like the helper to use a firm NO.

In conclusion, whether you like it or not, your dog will anticipate. It is a natural part of learning. Anticipate the anticipation and use it to your advantage.

~ Aamer Sachedina