1. Raising Criteria (successive approximation)

2. Breaking Down the Behaviour

3. Vary the Reinforcement

4. Relax Old Criteria When You Add New Criteria

So lets talk about,  5. Planning Ahead. It sounds very obvious but many of us have been caught off guard by our dog’s sudden leap ahead of what we had planned to work on. When our expectations are so much lower than what the dog has shown us he is capable of, we are then at a loss as to what to do next in the training program to keep the dog interested.

My suggestion is to have a goal behaviour in mind before you start a training session. Then break that behaviour down into as many parts as possible. Write them down.

The other option is if you have reached your goal behaviour very quickly, then raise your criteria through successive approximation. Essentially this means start to generalize the behaviour. Change location, ask for duration, or add distance to the behaviour to make it just a little bit more challenging.

He is a bit on the exuberant side so you should plan on stimulating his mind and body with training and lots of exercise. So far we have worked on Sit, Target, Focus (eye contact), and Leave-it. He has picked them all up very quickly, his brain is like a sponge. He is easily motivated and loves attention and praise.

If Tyson sounds like the right boy for you, keep in mind that you will get a free training session with me, when you adopt any dog from the SPCA in Newtown or Waikanae. This will help you get off on the right…paw! We can work on basic foundation behaviours, how to teach manners and listening around distractions! The session is for you to ask questions about any training issue that comes up, or simply to get you going in the right direction.

Call in to Newtown and ask to see Tyson, he is ready to find his life-long, loving home, hopefully before the end of February?!

A Reward Marker is a sound (usually a word or a ‘click’ if using a clicker) that means “Yes! that was right!”. This sound is followed by what is called a Primary Reinforcer.

Primary Reinforcer is naturally reinforcing. In other words a reinforcer that does not have to be learned. There are only four primary reinforcers, food/treat is the only one that is suitable to train with (the others are air, water and sex).

The Reward Marker (many people use the word ‘yes!’) is called a secondary or conditioned reinforcer. This is a reinforcer that the dog has learned it likes. The most efficient way of creating a secondary reinforcer is by using classical conditioning (associating the secondary reinforcer with the primary reinforcer).

Have I lost anyone yet? This is very basic learning theory terminology but can get confusing very quickly!

So lets say you have conditioned your Reward Marker as the word ‘yes’ or a ‘click’ noise from your clicker. This now means to the dog “I like what you are doing right now! You have earned a treat!” Now without ever having taught a cue or command the dog understand that when it hears the RM (click or yes!) it has done something right!

The benefits to understanding this small part of training are immense. Once you can communicate to your dog what he is doing that you approve of, he can start to work for that approval without bribing, commanding, or physical force. The more often you reinforce the behaviours you like (whether cues or automatic) the more often he will demonstrate those behaviours automatically. Automatically is important because that means he will be thinking about what he can do to get your approval, instead of what he can do to get your attention… jumping, barking, chewing, pawing, nipping etc…

Having your dog learn to do appropriate behaviours, on his own, will also make is easier to teach him cued behaviours (hand signal cues or verbal cues).

Another benefit to using a reward marker is the ability to ‘bridge’. Bridging makes it easier for the dog to understand what he has done right, in order to do that more often. I think this is best described in a scenario.

Scenario 1:  Dog sits, person notices dog sitting and because they haven’t installed an RM they say “GOOD DOG! what a good boy!” and they run to the kitchen to grab a treat.  By the time the dog gets the treat he has stood up and followed their human to the kitchen where they have gotten a reward for following the person to the kitchen.

If this scenario happens enough times the dog may learn that he has gotten the reward for sitting, or maybe he continues to think he is getting a reward for following. This process of learning can take a long time if the person uses the same words of encouragement  when the dog is doing nothing at all, or does not pair them with a primary reinforcer often enough for the dog to recognise the reward marker.

Scenario 2: Dog sits, person notices and says “Yes” as soon as the dog’s bottom touches the ground.  “Yes” in a happy tone of voice is this dog’s installed Reward Marker. The dog now understands that the moment he heard the word sit he was doing something the owner liked. Now he knows he has earned a treat. Follows the owner to the kitchen to get his earned reward.

The difference in the two scenarios may seem subtle, but the difference is in the speed of learning and understanding the dog acquires. A well installed and understood RM is the difference between teaching a dog a complex or new behaviour in a matter of minutes (or seconds) instead of days or weeks.

The Reward Marker is all about giving feedback. Feedback is invaluable when you have started a new job or when learning a new sport. Getting feedback from those you work with can be the difference between doing your job well and getting a promotion, and doing your job poorly (from lack of understanding what your boss wanted) and never moving up or worse, getting fired.

Don’t fire your dog for not knowing what you want! Learn to give feedback that your dog can use to get better at his job…. being your companion.

1. Raising Criteria (successive approximation)

2. Breaking Down the Behaviour

3. Vary the Reinforcement

So lets move on to “rule 4”…

4. Relax Old Criteria When You Add New Criteria: If you are working on a complicated behavior, with many parts, it is hard for the dog to maintain concentration on all parts at the same time.

I have recently begun learning to swim and it strikes me how similar swimming  (or any sport is) to training a dog to a multi-behaviour situation like “walking on lead”. Walking on lead has many different parts. We have already learned to break down those parts, to teach each part using successive approximation until that part is perfect, and to vary the reinforcement to keep the dog motivated. Rule 4 tells us that in order for the dog to be able to learn the full behaviour well, he has to be able to practice each part individually without worrying about the other aspects of the behaviour at the same time.

When I am taking a swimming lesson, we work on a number of different things. When I was first learning, we worked on kicking, while holding onto a floating board. We did this so I didn’t have to concentrate on anything else other than kicking. I didn’t have to worry about getting my breaths in, or what my stroke looked like, only kicking.

This helped me perfect my kicking (or at least improve it)! Then when we went back to strokes again, my kicking suffered a bit. But because I had practiced it alone, I was able to integrate the correct kicking with the stroke and the breathing until they were all passable. Now that they are passable, I go back to working on them individually again to improve them further.

The point is, to improve them I work on them separately then integrate them. You should do the same for your dog’s walking skills. Work on speed, checking in, heeling, and turns all separately. Then integrate, but don’t expect the behaviour to be as perfect, allow some time for adjustment and go back to individual practice regularly.

Goal Behaviour-   Leave –it: “Move away from that item and stop thinking about it.”

You can teach this behaviour by baiting your hand with a treat, and marking the moment that your dog moves away from the hand, then reward!

The dog gets the Reward Marker when he moves away from the hand. However the movement away was preceded by him first moving towards the hand, and possibly nudging it with his nose.

Superstitious Response- The dog continues to purposefully touch the hand, then back away thinking that the entire behavior chain is needed to elicit the Reward Marker!

How do you get rid of this response? It helps to have an established No-Reward Marker so that you can give your dog clear feedback about when he is doing a behavior incorrectly and needs to “try again”.

It will also help to generalize the behaviour. Practice the behaviour in as many locations and situations as you can think of, slowly increasing either the distance, duration or by adding low level distractions. As the dog gets his reinforcement even when he has not performed the superstitious behaviour he will slowly learn that the SB is unnecessary.

Superstitious Behavior Responses are even more likely in dogs who are not emotionally resilient. These are dogs that are more reactive and less able to calm themselves down after an environmental change. To these dogs routines are even more important and more significant than to your typical dog. Therefore they don’t generalize as quickly as other dogs, and as you noticed, generalizing is the way you prevent Superstitious Behavior and Accidental Reinforcement.

Accidental Reinforcement:

In this same category of “training issue” we also have the “Accidental Reinforcement”  AR most often happens when your own body language becomes a cue without you actually meaning it to! .. Example time…

Goal Behaviour-  Sit: “be in a position where you are upright with your bum on the ground.”

You will generally teach this behavior using a lure then fading the lure into a hand signal, then fading the hand signal into a verbal cue. Most of the time that will work fine, the problem shows itself when you attempt to use the verbal cue only. And although you have practiced, the dog does not seem to understand the cue. This can happen when your own subtle body language is actually what the dog is picking up on, and is not associating the behavior with your verbal command.

Maybe you step forward slightly when you lure, maybe you drop a shoulder, or raise an eye brow. The dog has picked up on this miniscule movement and decided that  was the important signal, because that was the most consistent signal! (Dogs may not generalize well, but they notice details VERY well)!

You can avoid accidental reinforcement by generalizing your dog’s learned behaviors, and by video taping yourself while training!

That’s right, there are many ways to motivate a dog (or another person for that matter) to do what you want. Rewards are a good motivator, but fear of pain or death is also highly motivating. Some dogs are so highly motivated to avoid pain, they learn very quickly from being shocked or choked into submission. The unfortunate consequences of this style of training is manifold. The biggest of which is if a dog does not respond to the training, becomes reactive or aggressive trying to defend itself or is so soft that the training effectively “breaks” them, they are considered “un-trainable”  and euthanized. Voilà! See, the aversive techniques have a 100% success rate because the dogs it didn’t work on are dead.

There are some lines of dog breeds that these techniques are particularly affective. K9 police dogs seem to be one of them. Most likely because of the reason mentioned above. These dogs are bred to tolerate many physical corrections without giving up or becoming aggressive to their handlers. The ones that don’t handle the training well are culled, either by re-homing or euthanized. This is survival of the fittest at its worst. The bright side however is that more recently police dog training has become a lot more positive as K9 trainers are realizing they can get better results, lose less dogs, and have a better K9-handler relationship with positive training methods.

I am hoping to see this first hand at the end of February when K9 trainer Steve White will be in Wellington for a two day tracking seminar! I have signed up my SPCA special, Ripley and me for the program and can’t wait! There still may be some spots available so come along if you can!

The first two blogs on this subject covered…

1. Raising Criteria (successive approximation)

2. Breaking Down the Behavior

So now lets think about rule number ‘3’.

3. Vary the Reinforcement: Use different levels of reward so the reward each time is either variable or differential (Random or better reward for better responses). Also use different types rewards; Praise, Petting, Food, Freedom, anything reinforcing to your dog! http://www.dancingdogs.co.nz/2010/07/things-my-dog-likes/

Why is it important to vary your reinforcement? Lets think about the way a mind is stimulated to continue doing a behaviour. Gambling is the best analogy. People become addicted to gambling, video games, and drugs for all the same reasons. It feels good. It feels good because of the SEEKING circuit in our brains. This pathway is a very adaptive part of the brain that enjoys searching, hunting, and achieving. When the brain is actively involved in these activities, it releases endorphins. Endorphins feel good! Gambling is a particularly good analogy because the brain is receiving endorphins from the SEEKING aspect of the slot machine/game, as well as the sense of achievement when you win! Slot machines are particularly addictive because they are set up on a “Variable Reinforcement Schedule”. This means that the reinforcement is random. You never know if you are about to win a little, lose completely, or get a Jackpot!These makes the game exciting and addictive.

If you were to use a slot machine that gave out a 20 cents for every 10 cents you put in, you would be making money… but how much fun would you be having? The very first distraction that came along and you would get up and leave that machine.

It is very important when training our dogs that we understand the use of reinforcement schedules and  pick the right one to use. The Variable (random) schedule is best to use when the dog is very fluent in the behaviour you are looking for. That way they have a lot of confidence that they are doing the behaviour correctly, and the become persistent with demonstrating the behaviour. In fact they try harder when they don’t know whether they will be getting one treat no treats or 5 treats because each behaviour MIGHT  mean a jackpot!

It is also very important to know that you shouldn’t use the same reinforcement all the time. Particularly because if the dog knows what is coming it makes it less exciting to win! So mix up those food rewards, use cheese, chicken, dog kibble, dog roll, and green peas if your dog likes them! But Also use petting, praise, and life rewards like being able to go sniff a favoured bush.

Varying your reinforcement will addict your dog to training. It will keep him wanting to continue to train as well as working hard for those rewards. But be warned, it doesn’t matter how many different types of rewards you use, if you are also using punishment to correct, as you will ruin the confidence and trust you are attempting to build. So remember if you get frustrated don’t blame the dog or the treats, instead go back to the basics and follow guidelines 1 and 2. Break down your behaviour further and increase the difficulty in small increments.

As we have discussed so far; to become a really good shaper the first rule is…

1. Raising Criteria (successive approximation)

which we covered in a previous blog.

So now you have been practicing raising your criteria for your goal behaviour in very small increments, and if you followed the previous blog, you should now have you dog walking at heel, at least in non distracting environments, off lead. So, time for step 2.

2. Breaking Down the Behaviour: breaking down is typically used for behaviour chains, or behaviours that won’t happen “automatically”.

Every behaviour you want, like walking on a loose lead, needs to be broken down into its constituent parts. And then each behaviour needs to be taught separately, so that you can be sure the dog is ‘fluent’ with each step.

Parts for Loose Leash Walking: Heel, Looking at handler, adjusting gait/speed to keep the leash loose, defaulting to heel when encountering a dog or a person on the walk.

If you think about it, the behaviours I have listed above can be broken down even further, making the learning process easier for your dog! As an example lets talk about defaulting to heel. Your dog is already fluent at finding the heel location, and likes to be there because it has a lot of positive associations, Its the ONLY place he gets rewarded while on a walk.

In order to default to heel he needs to be able to predict which situations will be rewarding if he chooses to heel. Which means, that every time he sees another dog, or a person, or a bicycle, or a skateboard, he should get rewarded. But only once you have lured him back to the heel position. He will quickly catch on that these objects predict that a food reward is available for him to receive IF he chooses to heel.

To help him along we have to make sure he actually sees the predictor, for this we can teach a “Look at that!” cue, or we could install a walk onto the grass and practice sits, and eye contact until the object had passed. 

The hardest part some people will have with this exercise is with dogs who are already reactive to certain moving objects  while they are on walks. For those dogs we need to make sure that we also work on a U-turn cue, an Escape cue (ducking behind a car or down a driveway), and always work the dog below threshold. These dogs will improve but the counter conditioning process may take a bit longer as you will be having to turn a  negative stressor into a positive predictor! As with all things prevention is easier than trying to fix something after it has become a habit, but it is still possible!

The trick with breaking down a behaviour is that you have to practice. It helps to write down your goal behaviour then try to figure out all of the physical/mechanical behaviours your dog will need to become fluent in before being able to achieve the goal behaviour. Then also think about what emotional state your dog needs to be in as well! Do you need  your dog to be excited, motivated, and focused for agility? Or do you need your dog, relaxed, calm, and passive for a relaxing stroll down a busy sidewalk in Plimmerton?

Then your goal is to teach each of these behaviours separately, make them fun and rewarding, then when your dog is fluent in each, start to tie them together!

Shaping is a term used by clicker trainers to describe how we influence and communicate to our dogs the behaviour we want, without using corrections. I know that sounds almost unbelievable. How does the dog know what they are doing wrong unless you tell them, right? Wrong! In fact if you learn to shape behaviours you are giving the dog tremendous feedback about what they are doing right. With this information the dog can then understand what it is you are looking for and figure out how to earn their reward! That reward can be treats, praise, or really anything the dog enjoys. However, when we are using the clicker to give feedback we always reward with food as that is a primary reinforcer and will be more memorable to the dog.

Ok so lets take this slow and use an example that one of my clients is working on right now! Helping their dog to be calmer while walking on lead and look to the handler and come in to heel, when another dog approaches. We start with the easiest part of this behaviour… finding the heel position.

1. Raising Criteria (successive approximation) : When teaching a cue you must raise criteria quickly but in small increments. Plan ahead. For maximum learning she should get it right 80% of the time.  Or raise criteria if she has demonstrated the behaviour perfectly 5X in a row!

If the first behaviour we are working on is heeling on cue, then we need to start easy. Take your dog to a safe off lead area, a big room in your house or your fenced yard would be perfect.

Start walking around the outside perimeter briskly, as soon as your dog comes to join you, click and reward her next to your leg (the reinforcement zone). Continue walking, if she continues to walk with you, click and reward her again! Try not to stop while you reward, but continue your brisk pace so she has to follow along with you to get her treat. Each time she leaves your side continue walking, if you notice that she has started to walk with you for more than one step, then it is time to raise criteria . You can raise the criteria for the exercise by holding off on the click until she has walked with you in the heel position for 2 steps. Click and reward every 2 steps, when she has done this 5 x in a row, hold off on the click until 3 steps. Continue to raise criteria until she is now walking by your side and you are only clicking and rewarding after each full perimeter walk!

Notes:

-Do not ‘tell her off’ or acknowledge in anyway if she leaves your side, simply keep walking and maybe even jog a few steps. Remember, you are competing with a  very interesting environment, we have tried to make the first part of this exercise as easy as possible by practicing first in a familiar area! You can also raise criteria by changing locations, but if you do this, lower your criteria back down to 1-2 steps and work back up to a full perimeter!

-If she never comes to your side, it could be because she has learned that it isn’t fun to walk next to people (this happens if dogs are harshly corrected when learning to heel on a lead). In this case click if she looks in your direction, you may even have to toss her the treat initially! But soon she will build up confidence enough to approach you, and when she does click and make sure she gets a very tasty treat (or two)!

See the thing is, I wouldn’t want to offer the same type of classes you can easily get from a puppy preschool or your local dog club. They would have to be more about how to help your dog learn and not just about how to get the perfect Down/Stay! I would want to cover shaping and clicker training for more advanced students, while still offering help with teaching the basic cues everyone wants their dogs to know… or do you?

What would YOU want from a dog training class? When you go to classes now with your dog, do you feel like you have learned exactly what you wanted and needed? Is there anything you feel like classes in general are missing?

I have talked to a lot of trainers and listened to a lot of seminars that cover would should be in a puppy or dog class, but I really want to hear what you would  most like to experience from a group training environment.

I have certainly had some feedback that a “Reactive Rover” style class would be a huge benefit to those of you who don’t feel like your dog can cope in the basic obedience classes due to over excitement or embarrassment that your dog is too distracting to the other participants. So that is obviously one that would be appreciated, but what else? Rally Obedience for those who like to earn ribbons but don’t like the fussiness of traditional obedience? Or how about a puppy class that teaches you and your puppy how to engage in appropriate play, rather than a free-for-all that seems to make behaviour problems worse?

If every single person were there just for you and your dog, what would you want from a group class environment?