Mr. Boing-Boing and the clicker

Training the heel.

Get out clicker. Get slab of carver's treats. Get mobbed by Two dobes and a Pinhead, all doing the dog equivalent of, "it's the CLICKER!! THE CLICKER!! My turn first. No, I want to be closer to her, back off! Look! I'm prettiest. Well I sat first. Ooooh, lifting paw here, admire me. I need food because I'm oldest.!"

Shoo two big black dogs into office with some difficulty, finally resorting to pitching a couple of pea-sized pieces of Carver's in then slamming the door to howls of protest.

Am left with Mr. Boing-Boing. "Oooo! It's me! Me, me, me!" (boing-boing-boing-boing-boing). Take tiny treat in my left palm, fold thumb over, hold hand down, two fingers extended downward, and start walking. Boing-whap, boing-whap, boing-whap as Mr. Boing-Boing smacks his head repeatedly against my fingers trying to get the treat. Finally, a momentary boing-trot. Click, treat in position. "ooo! Food!" Boing-whap, boing-whap, boing-whap, "hey, what's that on the floor?" looks down and floor and, amazingly, has all four feet on floor at the same time. Click.

"I get it! I get it! Look at floor, get a click!" Mr. Boing-Boing trots along, staring fixedly at floor, waiting for click. "What's wrong with you! I'm looking at the floor just like you wanted." Stares harder. Glances up at his stupid human to make sure she's watching how good he's looking at the floor. Click. "Okay, this must be it then," look at floor, look up. Look at floor, look up. Down, up. Down-up. Downup-downup-down-up. UP. I wait for two strides of looking up, click. Five more strides, looking up, perfect heel...CLICK!

"Good boy! We're done." "What do you mean we're done. I'm not ready to be done yet. I want more!" Boing-boing-boing-boing-boing! Pick up mid-boing and deposit into office, bringing out Cala for her session clicking for straight backups 90-lefts and come-fores from a stationary sit.

Clicker training makes me laugh.

Breakthroughs—Cala

Okay, compared to Ginger my month has been really boring. No intact girls, so nobody in season. No 16 dogs in the house, no wallet being stolen (thank goodness!) I have, however, made some nice training progress with both Cala and Zipper. I'll detail Zipper in a different post.

As I described in the blog on fighting fire with fire, Cala can be a very challenging dog to train. I've been following through with the toy game with her, and I'm thrilled that it's having really amazing results. She's listening and following direction. At the Columbia show she was able to handle Rally Excellent well and take the jump twice, both days, without screaming. Her dumb handler (that would be me) blew by a sign on Saturday so no Q there, but she got a 96 and a first place in her Excellent A class on Sunday.

Right now I'm working on refining some skills with Cala. Cala has difficulty with the halt, side-step right, halt. In that exercise the dog is sitting in heel position and the human steps directly to the right one step with their right leg, then brings the left leg over to meet it. The dog should get up and also step directly right (no forward motion). When we started, Cala was coming over to the right with her head and front but mostly leaving her butt where it was, so that she ended up sitting crooked and wide. We have progressed to her coming over with her front end then swinging her back end in to sit mostly straight. That's satisfactory for Rally standards, but I'd really like to see if I can get her to sidepass. What I would like to see is her getting up and moving both her front and rear to the right at the same time in one motion, then sitting down. And recently she's done this a couple of times with a fun leap to the right. So I'm marking that with a click and a reward.

Another problem spot for Cala are straight fronts, especially from heel position. In Rally signs #41 and #42, the dog halts and sits in heel position, then must move to front (sit in front of and facing the handler) without the handler moving. So the dog has to get up, move forward, and swing around to face the handler then sit again. (The other lower-level Rally call-fronts allow the handler to step backwards several steps to help the dog get straight, this one does not.) Cala either gets up then sits right back down or she gets up and sits sort of facing me but really crooked. She loves left (swing) finishes so she tends to sit at about 9:00 or 10:00 when I want her to sit at 12:00. I had been using my left hand, the one nearest her, to signal for the front. Recently I started switching to my right hand. I take the right hand with a treat in it, bring it to just in front of her nose, then tell her to front while simultaneously drawing that hand out in a big clockwise circle then back toward my right side. It's helping tremendously and her fronts are much better. This is a classic luring scenario, but I will be able to drop the lure quickly and just use my hand at a real Rally trial.

In agility, the riot stick game is working beautifully; when we can work. At the end of February Cala jammed a toe when we were training contacts. So she was off of agility for 3 weeks. In her first training session back I did a single jump, the aframe, and she came running back to me with a bloody foot. She'd cut a toe pad on a different toe. Sigh. So it's really just been this week that I've done a bit more agility training in preparation for the ASCA trial this weekend. I'll let you know how we do.

Clicking the heel

It's been awhile since I posted an actual training post. I've already written several on the theory behind the clicker as a tool for training, and writing something about actually applying the theory is long, long overdue.

I'm using the clicker to teach Zipper to heel. Zipper presents some special heeling issues because of his size. I want to teach an attention heel, but having him heel and watch me in the eyes will not work, because he's too small. To look in my face he would have to forge forward out of heel position and also turn sideways. So I need to teach him a different focus point. I am also a trainer who is more interested in attitude than in precision. Zipper will probably only compete in Rally Obedience, where half-point errors are not counted. I want him to think heeling is really fun and exciting and I want him to be somewhat accurate, but my goal is not a 200 score.

To that end, I'm using a technique I found on the Mighty Mite Dog Gear Forum, a news forum addressing issues (including training issues) specific to small dogs. I decided that I wanted Zipper to target my left hand in the heel, and that I would drop my hand down and keep it fairly still to act as said target.

Now Zipper already knows to target my hand with a nose touch. I needed something to help him learn to target on my left that I could fade. At first I tried carrying a toy balled up in my hand, and clicking and dropping it when he was in heel position. It turns out there were problems with this. First, it was difficult to get even a small toy bunched up in my hand, especially with a line attached. And if I didn't have a line attached, when I dropped the toy he'd just run off with it. Training session over. Second, Zipper is extremely toy motivated....sometimes. If he gets at all stressed he will not play with toys, and if he's distracted he also will not play. So toys weren't working. He's always food motivated, but food also had one problem; the frantically boinging Min Pin. I'd walk along and I'd have Mr. Ping Pong Ball. Boing!-Boing!-Boing!-Boing!-Boing! I could not get all four of his feet to stay on the ground long enough for me to click.

Enter the paint stick. It just so happened that when we last painted the dogwalk, we had a couple of those cheap wooden stir sticks left over. I dipped the bottom 2" of each stick in green paint to provide a color/smell contrast. Next, I started to click Zipper for putting his nose on the stick. I only clicked if he nose-touched the painted part. I knew that I would not want him actually touching the stick but rather following it, so the minute he understood that the stick was the target, I went ahead and put it in my left hand and began walking with it.

Wait. Back up. Let me describe that a bit because it's taken some juggling to get all the components right. I put the paint stick in my left hand, pointing down, flat side in front of Zipper. I also put the clicker in my left hand, flat against the paint stick, so that my thumb curls around and can click. I have to squeeze a bit to keep them both in there, but I don't want to tape the clicker to the stick, because I want to be able to slide the stick up over time. In my right hand are treats. When I click, I rapidly sweep my right hand over and treat in position.

At this point I've done three heeling sessions with Zipper, each about 5 minutes. We always start with Mr. Boing-Boing Dog, but by continuing to move forward and clicking when he's following the target with all four feet on the ground I'm already having progress. As of last night, he was moving forward and I was able to bring the stick up so it's no longer exactly in front of his nose but slightly above it. Over time, I will slowly withdraw the stick into my hand, so that my hand then becomes the target.

I also discovered another weakness in our training. Zipper thinks sits and downs are to be done facing me. I don't physically position Zipper (try to push or hold him in position with my hands) so I'm addressing this too with the clicker. At first every time I moved he'd get up. Again I realized a training weak spot--even when I've been working stays (very little so far) I've been doing so with him in front of me. So I worked a bit on me just moving back and forth, clicking when he stayed in position. I also worked on sliding sideways into him. If he accepted the slightest position that removed me from in front, I clicked. In essence, I have indeed taught him to watch my face when sitting and downing, so that's what he wants to do. I need to help him learn he can sit and down facing the same direction I am.

I'll update his progress over time.

Dog show weekend

I can't believe it's Thursday already. Time has definitely gotten past me this week.

Last weekend was the annual Columbia Kennel Club dog show. I hope some of you got to attend. There were many of us there, including quite a few of the instructors and all three owners. Kathy was chief bottle-washer and help-recruiter for Obedience and Rally. Ginger showed Ailsa in breed and Baker in Rally. Andrea showed Trip in breed. And I showed Cala in Rally.

I didn't know whether I'd be able to show or not; Cala had jammed a toe in agility practice the week before and obviously if she was still lame I wasn't going to ask her to jump, even the two low jumps she would have to negotiate in Rally Excellent. But that cleared up in enough time for me to get a bit of polishing-up practice done, so I felt we were ready. Or at least I thought we were ready until Saturday morning when I let myself get in a state over a bunch of stuff and only managed to walk the course twice (and even then I was anxious and upset). I ended up blowing right by a sign. There are many things that are forgivable in Rally. But not even attempting a sign? NQ. Not Qualified. Big fat zippo. Oh well. The good part was that Cala did the jump, both times, without going into a Cala-screaming-frenzy. This was really, really good news since I had thought I might have to retire her from Rally due to this loud overenthusiasm. Screaming like a demented banshee is perfectly okay in agility. In Rally, not so much. Very bad form. Missing signs was the order of the day. I was hardly the only one; in fact probably a majority of the competitors missed at least one sign. The judge had designed for one size ring only to get there and find it 10' shorter. So there were a load of signs jammed down on the end, and for some you had to go by one to get to the correct one. Lots of people missed or got lost.

Sunday I managed to get myself sane and we did finally get an Excellent leg.

This week seems to be medical week. Zipper went to get his teeth cleaned and x-rayed for submission to the OFA for hip dysplasia, elbow dysplasia, and legg-calve-perthes. Viva and Zipper both had acupuncture treatments today.

And my beloved Cala, on equipment last night for the first time since jamming her toe, took a chunk out of the pad of a different toe. We walked into the ring. She jumped two jumps and did the aframe, and came dashing back with a bloody foot. No vet for her, it'll heal. But big sigh nonetheless. Fortunately I only entered her in Jumpers at the ASCA trial the end of this month. Because it doesn't look like I'm going to be able to get much training in.

Capturing, Shaping, Modeling, Luring

Sounds like something a metrosexual would lisp provocatively on America's Next Top Model doesn't it? With Madonna's Vogue as the theme song. Believe it or not, Capturing, Shaping, Modeling and Luring are all dog training terms and describe ways we teach our dogs skills.

Capturing, Luring and Modeling are the easiest skills to learn. Shaping the most difficult but also the best way to really harness the power of the clicker. But what are they?

Luring
Luring is using food as, (duh) a lure. You stick a cookie in front of your dog's nose and use it to get the dog to do something, whether it's come (pull treat toward you) walk forward in heel (hold treat in left hand at nose level and walk), or sit (lift treat up and back so dog naturally wants to sit). Like a fish following a lure, the dog's body follows the dog's nose and the dog's nose is following the treat. Luring is easy to do but also very easy to abuse. It's really easy for the dog to expect that treat to be there and part of the picture. And because the dog is simply following his nose, he's not necessarily really thinking of what you want. He may not even notice he's now in a sit or following you, he's just following his nose. So lures should be very quickly faded into rewards. A lure is used to PRODUCE a behavior. A reward is used AFTER the dog offers the behavior.

Capturing
Capturing is pretty easy too, as long as you keep a clicker and small treats really close and handy. Capturing is when you just happen to catch a dog doing what you want and click for it. So you're sitting around watching TV and you have your clicker and some small treats you just carry in your pocket. Your dog doesn't like to down on command, but you know your dog knows how to down--all dogs sleep sometime! When your dog lays down, you click and pitch him a treat. Capturing is also a really great way to put fun stuff on command. Kathy has taught Kip to "commando crawl" by capturing. He likes to crawl forward on his belly, back legs dragging. She captured it by clicking it when he happened to be doing it. You can capture things like yawning, sneezing, begging—pretty much anything your dog does that you think it really cute.

Shaping
Shaping is more difficult but very powerful. Shaping is rewarding incremental steps to a final behavior. With shaping, you start global and go local. Shaping requires understanding your dog's body language and good timing, because you need to be able to click just at the right moment. Shaping also requires that the dog understands the clicker game, but dogs catch onto the clicker game really fast. The clicker game is that the dog knows that he can offer things and get a click/treat.

You can use shaping to, say, get a dog to pick up and hold a dumbbell. You can first click for the dog looking at the dumbell, then up the ante to having the dog touch it with his nose, then licking it, then taking it in his mouth for a moment, then holding it, then picking it up from the floor. All without pain or trauma, and the dog understands the lesson. Shirley Chong has a great detailed page on clicking the retrieve.

And here's a video of shaping from the Karen Pryor website.

Modeling
Modeling is the option least used by clicker trainers. Modeling uses physical positioning to teach. As an example, pushing a dog into a sit to teach the sit command. I try to avoid modeling if at all possible, because it's the slowest and least effective way to train since the dog is being physically put into position, which requires no brain power on his part and may be either scary or painful. It's always best to get the dog to freely offer a behavior rather than forcing the behavior on the dog.

--Robin

I'm Operant. How about you?

A lot of people who clicker train talk about having "Operant" dogs, or say they train "operant." And they're right. They are training using Operant Conditioning.

However, like I said in my other post, ALL dog training (and pretty much all people training too) is done using Operant techniques. Koehler, the Monks of New Skete, Diane Baumann, Bernie Brown, Karen Pryor and Sue Ailsby all use operant techniques, just based on different parts of the Operant spectrum. So a smart trainer understands the theory behind the foundations of our dog training.

Operant Conditioning means behaviors have consequences. And B.F. Skinner described four main quadrants of Operant training. It's really important to remember that the words we use for these quadrants are used with scientific definitions, not layman definitions. In other words, they don't mean what you might think they do. The four words are positive, negative, reinforcement, punishment.

Positive = adding something. (it does not mean pleasant, pain free, or happy)

Negative = subtracting something (it does not mean unpleasant or bad)

Reinforcement = increase repetition of a behavior (again, not necessarily pleasant, just an increase)

Punishment = decrease repetition of a behavior (does not mean painful or what I always think of, being spanked or hit)

The four quadrants of Operant Conditioning are:

Positive Reinforcement: Adding something to Increase a behavior.

Negative Reinforcement: Subtracting something (taking something away) to Increase a behavior

Positive Punishment: Adding something to Decrease a behavior

Negative Punishment: Subtracting something (taking something away) to Decrease a behavior

That's plenty of gobbledy-gook for this post. Suffice it to say, that for the most part, clicker training operates more in positive reinforcement and negative punishment than the other two quadrants. But all training does encompass all quadrants at some time or another.

—Robin

Click what?

Several of us at CCSC train using a method loosely called "clicker training." It's one of the training methods used in some of our classes, so I thought I'd start a bit of a conversation about clickers and what they are and what they can do. Before getting into the subject though, it should be pointed out that there are many positive training methods out there, and clicker training is just one of those. One of the fun things about training is exploring different methods to find ones that work best for you and your dog. As an example, though I clicker train, I also use voice rewards, toys, play, etc. in my training. It's all about having tools in the toolbox.

Clicker training as a subject is far too big for a single post (and besides, I know that the rest of our crew here on CCSC Dog Talk will also weigh in). But I'll start the ball rolling by describing some of the science behind the clicker. Hopefully you won't doze off, because it's actually pretty cool.

I think everybody has heard of Pavlov's dogs. You know, hungry dogs, salivating for dinner. A bell was rung before dinner was served and pretty darn quick, the dogs began salivating just at the sound of the bell, even when food wasn't present. The type of training that was called is Classical Conditioning. And the device, the bell, was a classically conditioned tool. The bell told the dogs to expect something--food. The bell meant that one thing and only that one thing. The dogs knew it, they understood it.

Well there was this other guy, too. By the name of B.F. Skinner. And back in the 1930s he published research into another type of learning, called Operant Conditioning. Basically, Operant Conditioning means that behavior has consequences. And all dog training and is based in some form of Operant conditioning. We humans also learn this way. I'm not going to go into detail about Operant Conditioning today. Just remember the term, because you'll hear it if you train dogs with a knowledgeable trainer.

The important thing to remember is that once a dog has learned that a certain sound always means food, he can learn that certain of his behaviors can have the consequence of getting to hear that sound; which will reward him with food. In other words, he learns to work for that sound.

Today we don't use a bell (they can clang when you don't want them to and don't fit real well in a pocket). We use a clicker. The clicker makes a sharp, audible sound. We use it to mark a behavior we want to reward the dog for doing. As an example, we could click the dog for sitting, or for watching us, or for a correct weave entry. Lots of things.

Why a clicker? Why not just praise?
Some trainers, including some of the instructors at CCSC, use only voice and are extremely successful. But clickers may be more effective than voice to mark a dog's success. Why? Because it's not voice. I know that sounds weird. But dogs have been bred for thousands of years to be exquisitely sensitive to our voice inflections and tone. And when we praise, we always put emotion into that praise. Unfortunately, the praise or "Yes" we say when we are happy and energized and training is going really well is often quite different than the "Yes" we say when we're tired and everything has gone wrong. It's just the way we humans are. And dogs read the real emotions behind the words. So when we grind that "yes" out with our teeth gritted, they know we still aren't happy.

The beauty of the clicker is that it is entirely black and white to the dog. It says one thing and only one thing. "This is what I wanted. You get a treat!" Because it's so unambiguous, and because it is divorced from emotion, dogs really get it. And because they understand it completely, the clicker becomes a tremendously powerful tool.

The second reason why clickers may work better than voice is timing. Studies have shown that we can click with our thumb (or finger) faster than we can process a verbal word, send it to our brain, have the brain send it to our vocal cords, and voice it. Dogs also process it much faster. Instead of having to figure out what word the human is saying, they get this brief noise with a very clear meaning. There's some evidence that clicking may connect directly to the dog's limbic, producing a stimulus response (see Karen Pryor link by clicking here).

Karen Pryor's site is a great one to start with if you want more information on clicker training.

—Robin