Getting the Behavior (so you can mark it)

So, now you want to train your dog to do SOMETHING... how do you get him to do what you'd like to train, so that you can reinforce it??

 

Getting the behavior (having the dog perform a certain behavior so that you have the opportunity to click it) can be done in several different ways:

you can simply wait for him to do something, and "click" it, then wait for him to repeat the behavior and click again, etc. (each offered behavior will come faster, as he begins to understand that the click actually means 2 things:

1)    he's earned a treat, because

2) he did it right! This is where the true power of this method lies -- it becomes a method of communication! With this method, you're teaching him that HE is the one who makes things happen (he is the Operant).  Another way of teaching the dog that he is the operant, is to have him perform already-known behaviors, and C/T them. Once your dog is "clicker savvy" -- once he understands that the click means he did it right -- you can't confuse him by clicking different behaviors, and you can click already known behaviors that already have "cues" or hand signals and commands linked to them, or unknown behaviors that you just think are cute, and when they become more "solid", then you "pair them"  with a cue.

 It sounds complicated, but it's not -- and the dogs LOVE this stuff -- training becomes a GAME!!  It's easier for humans to learn if they see it in action, but you'll learn it quickly too.

 

**Remember this simple rule: never click without treating, and never treat without clicking!**

Doing either of these will lessen the power of the clicker as the secondary reinforcer -- if you hand out treats sometimes without clicking, then why would the dog pay attention to the click? By the same token, if you click without treating, you're breaking the bargain -- the click means he's earned the treat.

4 Methods for "Getting the Behavior"

This part is a little important, because there can be little pitfalls built into this for those of us used to using force-based methodology in training... we're used to moving them into position, because most of us just don't have the patience for the purer forms, like shaping or capturing.  It can feel like an eternity waiting for your dog to offer a specific behavior, just so you can click it... then wait for it to be offered again...

There are benefits and problems associated with each of these methods:

Luring -- using the reward to get the dog to follow a movement. This is probably the most-used of the methods -- it's fairly easy to teach a dog the "sit" or "down" using luring. If your dog doesn't "sit" or "down" on the verbal cue only, it may be because you started out using a combination of verbal cues and hand signals. Dogs tend to assimilate the hand signals much more quickly, so if you're luring, they're dependent on the hand signal. There's nothing wrong with that -- just some people prefer the verbal cue. My dogs work from hand signals (which I think is pretty cool, actually!) because I depend pretty heavily on luring.

Capturing -- waiting for a particular movement. This is the "purest method", totally reliant on the dog. The problem is that you may spend huge amounts of time, waiting for a particular movement.

Shaping -- setting the dog up to perform a particular movement. This is mainly used to "add-on" to behaviors -- you can either wait to see what the dog will do on its own, or use some highly creative and imaginative ways to cause him to do what you want.

Modeling -- placing the dog in position. This is the least desirable method of "getting the behavior", as dogs tend to rely heavily on the physical handling aspect -- once the behavior is in place, it can be difficult to "phase out" the necessity for placement.

I personally use luring and shaping most of the time... they just feel more efficient to me.  My patience is good.... but not never-ending. lol

Shaping the Absence of Behavior

One of the hardest concepts for owners to learn when starting clicker training is "shaping the absence" of behavior... this is because we're taught all our lives to focus on what others are doing wrong, instead of what they're doing right. Changing this focus can be incredibly difficult... it's very difficult to ignore behavior, when you've been conditioned for your entire life to focus on it!

Note: for either of the following 2 exercises, you need a HUMAN partner to work with: one will be the trainer, one the trainee.  You'll each learn how and why positive reinforcement works the way it does.

Exercise 1: Focusing on Good Behavior

To illustrate just how difficult it can be to make this change (and, to give yourself some practice at it), have the trainee stand in the middle of the room. You're going to reinforce the trainee for doing anything that approximates a behavior that you have in your head... but the only thing you're allowed to say is "yes". Don't use body language, expression, writing, or any other way of communicating.

Exercise # 2:

do this exercise after the first, to provide contrast in teaching methodology. Think of another behavior for the trainee to perform. This time, though, say "yes" when you want to mark an approximation, and "NO" (loudly, to approximate a correction with a choke collar) when the behavior does not.

After completing both exercises, ask the trainee to describe how they felt about each of the methodologies. Most people admit that they feel "okay" about the first... but that the second made them experience frustration, resentment, and even rage at the trainer. Just something to consider, in making choices. Use of shaping the absence of unwanted behavior is an incredibly powerful tool that you can have at your disposal... I use this to reduce all sorts of inappropriate (by human terms {grin}) behaviors!

To illustrate how efficient this can be, I'm going to apply what may appear to be "circular logic"... you'll soon find out, though, that this is one of the few instances where this type of logic actually is applicable. {grin} Understanding how "shaping the absence" works will help you to become much more effective in your attempts to reduce and eliminate behavior that you don't like!

Again, reinforcement drives behavior. Always. So, reinforcing appropriate behavior will increase the instances of that behavior being seen again.

For example, if I consistently reinforce a dog for sitting to greet me, and that dog sees greeting me as reinforcement, then sitting will increase in frequency. I've just reduced the frequency (probability) of jumping behavior, by teaching the replacement.

But, we can take this even further than replacement of undesirable behavior... by shaping the absence of the undesirable behavior. Using the jumping dog above, I simply start to mark and reinforce (click and treat, if using a clicker) any behavior that isn't jumping, and that I deem acceptable as an alternative to the jumping behavior.

This means that, in greeting people, or in any circumstance that has predicted in the past to cause this dog to jump, I will reinforce

1. sitting to greet

2. standing to greet

3. lying down to greet

even if these behaviors appear to be accidental.

How does this help to reduce inappropriate behavior? By giving your dog options that he knows have been reinforced in the past... it's all about probabilities. The more appropriate options you provide (by reinforcing them), the smaller the probability of the inappropriate behavior being offered. If, for example, the dog only has 2 behaviors to choose from (sitting or jumping to greet), then you might end up with a 50/50 split -- the dog is still jumping half the time. But, if you also reinforce standing and lying down to greet, then you may well end up with a dog who performs these 3 appropriate behaviors in addition to jumping -- so that you'll likely get a 75/25 split... the dog is only jumping 25% of the time, now. AND, he's only jumping 25% of the time if he's being reinforced for jumping... by consistently removing the opportunity for reinforcement of jumping, it's quite possible to completely remove that behavior from his repertoire entirely!!

I use "shaping the absence" of behavior all the time, in working with aggressive dogs (for example, I've been known to reinforce a dog for scratching, rather than aggressing!)... it's an incredibly powerful tool, and one that I think all owners should work to incorporate into their teaching "toolbox"!

Teaching Redirection

 

What is redirection? Redirection means teaching the animal to perform a different behavior as a means of obtaining reinforcement -- a replacement behavior. A replacement behavior is, simply, a behavior that is taught as a substitution for a less desirable behavior. This technique can be used for most minor behavioral problems, and can be incredibly simple to teach, with a little ingenuity. In order for this technique to work, the replacement behavior must either be

1) as reinforcing as the replaced behavior, or

2) paired with a reward that is just as reinforcing as the replaced behavior.

Question from a Reader: >>In my particular circumstance (dogs alarm-barking at living room window when someone passes on the street) how do I teach redirection?<<

First, I would stand with him at the window when NOTHING of interest is going on out there. Reward calm behavior using The Most Boring Treats that you can use, and still have him eat them. {grin} You can use a cue with this, but I tend to stay away from that, in situations where you want a new behavior to be the *default* behavior... it just seems easier to me, for the dog to learn to default if the behavior doesn't *require* a cue.

Do this step at least 3x per day for 3 days. Then try to set up exercises using stimuli that cause him to react on a low level... it'll help if you can list 10 things that cause him to react, in order from lowest reaction to highest level of reaction. If the lowest level is a bird outside the window, for example {grin}, have the bird sit across the road (distance helps to lower the reaction!), and HEAVILY reinforce Max for looking at you -- even if he then immediately goes back to barking. The goal here is to get a foot in the door, training-wise... what you're doing is exchanging one "evil" for another -- you're turning the KILL-bark into a HAPPY bark (classical conditioning: forming a Conditioned Emotional Response to the trigger).

Then, it will be necessary to teach redirection. If you can't find a distance where you can get that initial "look" when you're at the window, then move to working outside. Dogs often show a completely different response to things when you change location. Do this 3 times a day until he's able to SIT while looking out the window -- even if still heightened. Barking or not is optional, at this point. If you want non-barking, then incorporate that criteria, too, at this level.  You can speed things up here by incorporating Jack-pot reinforcement for those trials that more closely approximate the end behavior you’re looking for… if, every 20 or so trials, he looks at the bird while standing WITHOUT barking, then that’s a Jack-pot.  I would jack-pot this until he’s at about 50% regularity, then I’d make this response level the bar (although you can use a lessor-value reward, including verbal marker only, to mark this… no reinforcement unless the desired behavior includes “no barking”.  You can keep the jack-pot by raising the criteria here (use your imagination… have a neighbor walk past to elicit responses from him, and shape the behavior you want.  In this instance, I’d go for a happy bark as the human starts up the drive (remember: successive approximation is the key here),

Then, start standing 1 foot behind him during the exercise, and start again without distractions. While he's calm, take him to the window, have him sit, then back up a step. Have him come TO you, then reward. Do this for 3 days, at least 3x daily. Then, add that lowest-level stimulus again (the bird, remember? lol), across the street. Reward the HELL out of him for coming to you. If he turns his head toward you, he gets ONE piece of the reward... coming to you gets FIVE pieces. Then, work at further distances (successive approximation exercises), and with each stimulus  (trigger) in your list of things that cause him to react. CAUTION {grin}: Be careful what you're reinforcing, in using redirection. Redirection is a WONDERFUL thing... but the SEQUENCE is EVERYTHING.

Here are 2 examples, to illustrate: Dog nips, give toy. You've reinforced NIPPING. Dog nips, cue "sit", give toy. You've reinforced SITTING. Watch the sequence... if the frequency of unwanted behavior escalates, change the sequence of the exercise. Just a note, because I see these mistakes with owners all the time (and, I may actually own the patent on them... I'm pretty sure I have that paperwork somewhere! lol)

 

A dog who barks in reaction is handled most effectively through redirection... in the natural progression, it's important to first teach the dog to be HAPPY to see the "trigger" (which may mean that the reactive barking will simply change to HAPPY barking -- but that's okay!) It's important for our dogs to understand that it's okay to "act up" a little, so that they won't be wary of trying new things. That's part of what you'll be teaching your dog, in using redirection... that he's not going to get in trouble for doing his Doggie Stuff. At the same time, you'll be teaching him to Lighten Up when presented with a trigger. Now, though, you need to reach a *compromise* -- and that's really all it is... you're going to teach him that it's okay to *alarm* bark, but once that alarm has been sounded, then it's time to turn it OFF. He'll get it, and so will you -- I have every confidence. At this point, you’ll teach him that instead of this sequence: Bark ----> Feed (Which taught him to associate Good Things with the TRIGGERS (So now he uses his higher-pitched HAPPY bark, instead of that evil-sounding KILLER bark {grin}) You'll use this sequence: Bark ---> "Quiet" ---> Stop Barking ---> Feed

This will allow him to STAY happy about the triggers, but also give you a way to turn it OFF -- Everyone's Happy. {grin} In effect, you're first making a Changed Emotional Response (classical conditioning, for all you techies), then switching to Operant Conditioning in the end, to make him aware that his behavior has consequences associated with it. You'll drop back to the classical stuff when the dog looks scared or upset, or when he's REALLY barking aggressively at a trigger. (Remembering that it's always best to not allow those situations... but that sometimes, life just HAPPENS). The point of the classical stuff is to affect a Changed Emotional Response, remember? Once you've got that  change, you move to operant -- and you're home free!

Here are some simple replacements for some annoying behaviors: Play-barking at other dogs -- replace this with allowing your dog to move toward the other dog. If your dog is play-barking at other dogs, teaching this replacement can be incredibly reinforcing -- as long as he's barking, he doesn't move forward. When he stops, he'll get closer to playing with the other dog. (You may need to practice this several times, but he'll get it!) Set him up with one other dog, both on-leash (This will only work if that “other dog” isn’t trying to kill him). Walk the dogs past one another, at tolerable distances, to start. A tolerable distance is the distance where your normally barking dog shows interest only. Walk him past this other dog, from one point to another, treating for appropriate behavior, until he's relaxed and happy. Then, move a little closer. When he can pass within 3 feet of this other dog without barking, find another dog to proof him against.

Note: it often works well to start an exercise with the first dog (it might take him a few reps to remind him of the rules of the game), then move to that second dog… I think you’ll see improvement quicker this way,  because he’s in the behavior groove now.

Jumping on people -- this is done to gain the reward of attention from other people -- use this to your advantage! Make it a rule that, if the dog is jumping on people or otherwise being a nuisance, he gets no attention -- when he's sitting patiently (or even impatiently!) he gets what he wants!

You can set this up the same way as the previous "Barking at dogs" example, using people. It is important that the people know what to expect, and how to respond, up front (I once had someone use the knee-to-the-chest procedure on my Saint pup -- I was pretty well P.O.'d about it -- explain proper procedure!!) When all four feet are on the floor (and his butt, if that's what you want) then he gets profuse treats and attention -- very simple, and it works!!

"Petting" people -- I include this, because I have a dog that does this. This tends to be a breed-specific behavior, and several of these breeds tend to be large and giant breeds! This is one of those behaviors that is cute when the dog is a puppy… not so much when he/she is getting nigh to 100 pounds! My dog doesn't do it as frequently as she used to, but she will revert to this behavior, if she's not getting her share of the petting. This can be incredibly painful, coming from a dog this size! As a replacement, I've taught her to "say howdy!", which means to sit and shake hands. Now, I can have a person come in, tell Cis to "say howdy!", and she'll sit and offer a paw in greeting. Very simple. Your imagination, and those things that motivate your dog, are your only limitations when using this technique. With a little creativity, I'll bet that solutions can be found for just about ANY undesirable behavior!

So, What are Rewards?

Positive Reinforcement is anything occurring immediately after or during a behavior, which increases the chances of that behavior being offered again. A behavior that is rewarded is repeated.

For instance, if you call your dog to you, and give it a piece of cheese when it gets there, you've increased the chances of the dog coming to you, the next time. If you only pet the dog, or do nothing when it gets to you, the behavior may very well extinguish itself on its own -- the reward must be reinforcement for the behavior -- meaning that the dog may stop coming when called (petting and praise, while considered to be positive in nature, are very limited forms of reinforcement -- when compared to food, toys, or other possible motivators -- such as other dogs).

Always keep in mind this simple rule: whatever reward you are using MUST be more tempting than any distraction, in order to work. If the dog is more interested in chasing the rabbit than in coming to you for the piece of cheese, don't allow him to run free (unless he's earned by doing something spectacular... once you've done this, there is nothing your dog won't do for this type of opportunity!)

Comment from a reader: If my dog has decided that she wants a different treat from the one I'm offering, she'll take the biscuit and drop it and wait for another treat...... {grin} She just knows that there's better stuff to be had... maybe she thinks that what she's doing is *worth* more than that biscuit. I know that I've (on occasion) thought to myself "Boy, I'm not being paid enough to deal with this!" lol

One of the biggest concerns from correction-based trainers and cross-over handlers is that the dog will learn to only do what you want when there are treats available. Let me put your mind at ease: my dogs perform at a very *reasonable* rate of success, without food -- the clicker and food are TOOLS used for teaching new behaviors, and for "brushing up" on known behaviors.

The difference between using the clicker/food training and correction collar training is in the *fallout* that's inherent in the method... the correction collar is simply a TOOL, just the same as the clicker/food. Owners and trainers who were raised on dominance theory are also often concerned that, in using these methods, the dog is "training them"... in effect, they become "the treat dispenser". This, too, is true (if you choose to look at it this way)... because training is a bargain. It takes a strong person to let go of those old beliefs and concerns.

Brenda Rushman CCBC

 
 
 

Help!