My Operant Dog!

Enjoy the carnival!

This post is part of the second Assistance Dog Blog Carnival, hosted by L^2 (L-squared) at Dog’s Eye View. This month, the theme is “Decisions.”

**Breaking News: The Carnival is up, and it is huge! 26 bloggers participated this time. Please visit the carnival and enjoy!**

I can’t think of a topic that is more suited to a discussion of assistance dogs. The life of a service dog (SD) partnership is one of constant decision-making. For people, there are the “big decisions”: whether, when, where, or how to get a SD; which methods to use to train or handle the dog; which skills to train; how to deal with public access issues (when to confront, when to avoid, when to gently educate); when to retire our working partner and/or when to take on a successor, to name just a few.

In daily life, both members of the team make decisions all day long. On the human end of the leash, we decide how to direct our dog, which behavior to ask for, how to respond to a missed cue, how best to explain to the dog what we want, when to give the dog a break and when to keep working.

On the canine end of the leash, the dog is making decisions: When to alert, how best to guide around an obstacle, whether to use paw or nose, whether to use one paw or two, whether to refuse a command.

But even before the dog becomes a full-fledged SD, he is making choices. And the trainer is influenced by those choices, herself. That’s what my post is about — a recent event where Barnum’s decision-making shone through, and how that moved me to make up my mind.

I use the phrase, “The other end of the leash,” above, intentionally. My decision about what to write for this carnival post was inspired by a recent blog by behaviorist, and author of The Other End of the Leash, Patricia McConnell.

In her New Year’s blog, McConnell presents the difference between “resolutions” and “commitments.” She reveals past and current commitments she has made to herself and to her dogs on New Year’s and invites readers to do the same.

She provides terrific advice on how best to keep our “New Year’s Commitments” to ourselves and to our dogs:

We know that commitments are most often kept if 1) they are focused and specific, 2) they are attainable and 3) they are made public. Not that you have to blog about them yourself, but …. do what you can to put them out into the universe in some way.

McConnell also delivers a fascinating etymological tidbit:

The word “decision” originated from “decis,” or to “cut off” — as in, “cut off all other options”. Truly making a decision – truly – means selecting one option and completely eliminating all others.

I love words, so this intrigued me, but it didn’t seem particularly useful at the time. Instead, I got caught up in the idea of New Year’s resolutions for our dogs — her resolution for her dog, Willie, as well as the dozens of comments by readers who followed McConnell’s lead in making promises for the betterment of their dogs in the coming year — making it public by posting on her blog.

I don’t normally make New Year’s resolutions, but the more I read, the more I started to ponder, “What is a doable goal that would benefit Barnum and me, that I don’t already have in my sights?”

Here’s what I came up with and posted in the comments section. (I had so much trouble getting specific that I had to post twice):

[My first comment:] My resolution — and I wouldn’t have made one if I hadn’t read your blog, so thank you — is to teach Barnum an actual TRICK. Not a foundation behavior for service work or manners or obedience or handling, etc., but a totally useless, just-for-fun trick.

[My second comment:] Oh, shoot, that wasn’t very specific. I’m not sure what we’ll do yet. Maybe twirling? I want to see what he likes.

How about, at least once a week, we will train some sort of completely useless behavior, just for fun.

Why, when we have so much to learn that is crucial — assistance skills, obedience behaviors, manners appropriate to working in public — have I decided to waste time training something I characterize as “useless“? The question itself provides the answer: Barnum and I have so much pressure on us to succeed, we need to be able to do something simply for the pleasure of training. This will also strengthen our bond.

In fact, Karen Pryor‘s book, Getting Started: Clicker Training Your Dog, suggests beginning with a shaped trick — something for which the outcome is not important. That way, both dog and person can simply focus on learning and having fun, without getting weighed down by the need to “get it right.”

For people training their pets, this is important advice. If you’ve turned to training your dog to correct a problem, such as jumping on guests, counter surfing, or not coming when called, you don’t want to claw your way up the learning curve simultaneous with trying to fix an “issue.” You learn clicker by doing; it’s better to learn to teach your dog to “sit pretty” or balance a biscuit on her nose, and then tackle “jumping on guests.”

As a disabled handler training my own service dog, the pressure to train — virtually constantly — and to do it all perfectly, is multiplied by twenty. Barnum must learn all the good manners a well-behaved pet needs, plus cast-iron obedience, dozens of complicated service skills, fantastic impulse control, and to perform all these behaviors in distracting, public environments.

And who has to mold this wild adolescent into a top-notch helper and public ambassador for assistance dogs? Me! The person who needs the help!

Clearly, therefore, a great deal of pressure comes from my needs that are currently going unmet, and my hopes and expectations that someday, this will change. I remember what it was like to have these needs met by Gadget.

While Barnum gives me a great deal of love and enjoyment, there’s no denying that, for the past year, and the one to come, he also requires tons of work, without providing practical assistance in exchange.

Usually, I’m at peace with this uncertainty of when, or even if, he will start giving back. Occasionally, though, when I’m extremely ill, I think, “Can’t you just pick up that pen? It’s really not that hard.”

Or, “Dude, can’t you just open the door yourself? I’m in too much pain to move.”

I have my doubts: “Will we ever get there?”

Then, I have to give myself a mental shake and remind myself of who Barnum is and how much he still has to learn. It is this pressure to get my own needs met, that runs counter to Barnum’s needs, which I must resist — for both of our sakes.

A bit of additional pressure — completely unintentional — comes from external forces, such as from other trainers with whom I communicate, as well as from some of the people in my life. For example, when Barnum was about four months old, one of my PCAs asked when he’d be able to bring me water from the refrigerator. I was speechless, because this task was so far beyond our grasp.

My helper followed up with, “Within two months, do you think?”

She was not joking. She had seen Gadget working as a fully fledged SD. She didn’t see the years of effort that went into his training. She didn’t realize she was asking about a behavior chain that required the mastery of several discrete skills, strung together, combined with distance, distraction, and duration.

I think I said something like, “No, that is a long way off. First, we have to finish toilet training!”

Thus, my resolution to teach an honest-to-goodness, good-for-nothin’ trick!

You may have noticed I said we would do a “shaped” trick. So, what is “shaping”? What does it mean to say a skill is “shaped?”^[1]

As I mentioned in one of my earlier blogs, clicker training relies on the laws of operant conditioning, specifically, positive reinforcement. (We’ll come back to that word, “operant,” in a little bit.)

Shaping, sometimes referred to as “free shaping,” is considered the most advanced form of clicker training because there is no prompting by the trainer. Instead, we use a dog’s offered behaviors and reward those that resemble — in tiny ways, at first — the end result we want. The dog has to do more thinking than in any other form of training. It is a step-by-step way for dog and trainer to problem-solve their way to a solution.

The first “trick” I shaped with Barnum was to file his nails by scratching them on a board covered with sandpaper. (To teach your dog to do this, visit trainer Shirley Chong’s nail file instructions.)

In the beginning, this meant just clicking and treating (c/t) for looking at the board, then moving toward it, then putting a paw on it, then moving his paw. Over time I start refine the behavior (giving it “shape”): I wait for a paw movement that is a tiny bit like a scratch — the front of the paw is bent forward, or he moves the paw backward as he brings it down. I stop clicking for just “paws on board” and only click for raking movements.

Eventually, I refine more. At different times, I have shaped for scratching only the sandpapered parts of the board, for longer scratches, harder scratches, and for two or three scratches in a row. Barnum also seems to be “left-pawed,” so I click more for right-paw movements to even him out.

This behavior is not yet finished, but here is a three-minute video of one of our early training sessions. This was in October, when Barnum was six months old (and looking like a very leggy, shorn black sheep!).

Note: I haven’t captioned the video, because there is no sound other than clicking, and there are so many clicks, the software is not refined enough to get the timing right. However, you can see when I’ve clicked, because Barnum looks up at me, and I throw a treat.

I also have not provided a transcript because, again, there is no dialogue, and my guess is that reading a transcript would be tedious.

But, here’s what occurs: There is a plywood board about two-thirds covered in sandpaper on the living room floor. Barnum immediately goes to the board when it’s placed on the ground and rakes a paw across it.

Initially, I click for every scratch with one paw. After a while, he ends up on the end of the board where there’s no sandpaper. When he scratches there, about three times, I withhold clicks, because we’ve already worked on this. Then he remembers, “Oh yeah, I don’t get paid over here,” and repositions.

Then, I start holding off on some clicks to get two scratches. I click even if the second scratch is very abbreviated, because my criterion is just “more consecutive scratches.” I also click more often for one scratch if it’s the right paw, even if it’s not a very good scratch.

I clicked about 25 times in this session (which is actually a very low rate of reinforcement [RR]. Higher rates are much more effective, but I’ll save that discussion for another post). However, the biggest error I made is that I went too long. Barnum tells me this quite directly by wandering away at the end of the video. His poor puppy brain was full!

Clicker sessions should be short; shaping sessions — because they are so mentally intensive — should be particularly short; and given Barnum’s age, it should have been even shorter. A one-minute or 30-second session would have been ideal.

What I hope you notice most, however, is not what I did, but what Barnum did, which is ultimately what this post is about. Through clicker training, Barnum is learning to make choices. He’s learned he’s free to guess and investigate: Will this work? No, then what about this?

Since there’s no punishment, there’s no reason not to try something. Novel behavior is usually rewarded, in fact.

In other words, it’s obvious when watching this video that Barnum was thinking. Sometimes he pauses and tries this or that part of the board, including swinging his body around to try different angles. Sometimes you can see him deciding to use one paw instead of the other, especially when the first paw didn’t get clicked. And, most importantly, he didn’t give up — he made the decision to keep trying until he got too mentally tired, and took a break. Also a valid choice, and an important communication to me.

It’s worth noting, however, that — although you don’t see this because it occurs after the tape ends — after turning and walking away, Barnum arced back around and immediately tried to continue. In fact, this “game” is so reinforcing for him that even when I have gone too long, and he clearly needs to stop, when I pick up the board, he always tries to get more scratches in. I will be lifting up the board, and he’s trying to stand on it (and that’s 80 pounds of dog — oof!).

That’s where we were six months ago.

My, how things have changed!

At that time, this skill was the only one I had trained entirely by shaping. (In most cases, I’d used a combination of shaping and targeting.) Since then, Barnum has gotten more “clicker savvy,” and we do more pure shaping. As a result, he is much more accustomed to offering behaviors.

I didn’t realize how much learning was going on inside his furry head until I embarked on our first twirling lesson. Remember that New Year’s resolution? Our useless trick?

Here’s how it went. I got out my treats (cubes of frozen beef heart), and my clicker, and shut us in the larger bathroom — the room with the fewest distractions that still provides enough space to maneuver.

My first decision was which direction I was going to teach him to spin, and more importantly, how I would remember which direction I chose. Shaping requires split-second timing; a good shaping session is a series of reactions without pauses for thought.

In fact, since no response (the absence of c/t) means, “That’s not what I’m looking for, try something else,” if I stop to think, “Now should I click that or not? Was that clockwise or counterclockwise?” I very well might be giving Barnum bogus information — telling him he didn’t do the desired behavior when he did.

Therefore, I needed a way I could keep track of which direction I’d spin Barnum without my brain inhibiting the speed and rhythm of my hands. Barnum has a streak of silver on his left, front leg. I decided to shape him to spin left, using that visual landmark to keep me on track if I became confused.

Now it was time to tell Barnum the plan.

I sat and waited. When Barnum looked to the left, c/t. I cheated slightly, by including an element of luring — I tried to throw the treats to his left. This meant, every time he turned toward a treat, I could click him again for that left-turning motion. We went on like this, with him making big, slow circles as I clicked and lobbed meat.

Did he know what he was being clicked for? No, not yet. But just like an athlete or pianist builds muscle memory by practicing moving their body in a certain way over and over, I had faith he would eventually start turning to the left on his own. At first, he wouldn’t realize why he was doing this. Eventually, he would have “a lightbulb moment” and start doing it intentionally. Then, I would shape a tighter, faster spin — maybe. Someday?

But for now, I was just trying to get him moving in the correct direction, literally. All was going well until, for whatever reason (probably due to where I’d thrown the last treat), I didn’t have anything to click. Barnum just stood perfectly still, watching me.

When no c/t was forthcoming, he offered a sit, because that is the behavior he has the longest reinforcement history with. I continued to wait. I figured eventually he would at least look left, I’d reinforce that, and away we’d go.

Instead, when he didn’t get clicked after a few seconds, he pawed the ground! First with one paw, then the other. He tried both tapping the ground and raking it. If I’d been prepared and thinking on my toes, I could have clicked one of these left-paw movements, but I was not expecting it.

I had been shaping him to hit a target on the floor with his paw, and of course he knew the scratches from the nail file, but there was no target or filing board this time, he was just trying out different foot behaviors — maybe partly because he recognized that the behavior had something to do with moving his legs.

Then he sat again. No click. He downed. No click. He did all of this facing me, straight on, so there was no clickable movement to the left!

He lay there, looking at me. Then he elaborated on his down. He put his head on his right leg. Then his left. He tried rolling partway onto his side. (Again, looking back, I realize I could have clicked him putting his head on his left leg, but I was so entranced by him offering me behavior after behavior, I wasn’t thinking small enough.)

Then he pulled himself back into a sit. He locked eyes with me and stared — eye contact, a behavior we work on every day.

Up until now, every behavior he’d offered was something he’d been rewarded for in the past. (The head-down-on-paws and laying on his side are some of the “bio-feedback” behaviors I click for when asking him to relax.)

When nothing in his known repertoire worked, he did something that totally surprised me. Staying in his sit, he scooted his butt a couple of inches to one side. Then he tried scooting the other direction.

I have never seen him do this before or since. I have certainly never asked for the behavior, “Sit, looking at me, and — staying in your sit — move two inches to the right.”

He was getting quite frustrated. He wanted the game to start again! He shifted his eyes left (probably a sign of mild stress, actually), and I clicked! He startled (“Hooray!”),  jumped up, and dove for his treat. We were off again on shaping the twirl. I ended soon after.

To someone who is not obsessed involved with clicker training, it might seem strange that I was thrilled that Barnum offered me a whole bunch of behaviors that I did not ask for. That’s because traditionally people think of dog training as meaning I tell the dog what to do, and if he “obeys,” he’s a good dog (and I’m a good trainer). If he doesn’t, he gets a verbal reprimand or a leash pop.

However, my goal since Day One has been for Barnum to be “an operant dog” which is clicker-trainer slang for a dog who has learned that thinking and experimenting, making many choices, will not only not be punished, but will be reinforcing in two ways: it will be fun in its own right, because it becomes a game to puzzle out, and because he can win this game over and over, by receiving positive reinforcements (usually food, but occasionally play, toys, physical affection, freedom, or other experiences he values).

An “operant dog” in the clicker world is one who offers behaviors without waiting for a cue, target, or lure. The dog is not passive or reactive, but instead is operating on the environment, making choices.

As a SD, Barnum will need to learn very complex skills, and this will be made much easier if he is able to problem-solve on his own. I’m currently laying the foundation for a service skill that will require him to “throw multiple behaviors at me.”

The task has to do with my sometimes falling asleep while I’m infusing intravenous medication. The IV pump has an alarm on it, but I have sleep disorders for which I take several herbal and pharmaceutical sleep aids. As a result, I frequently wake up just enough to turn off the alarm and then fall back asleep. I awaken many hours later, still hooked to the pump, and not having flushed my line — with no memory of shutting off the alarm.

I want Barnum to wake me up and keep me awake until I disconnect and flush my PICC line. I began training this skill simply by making him aware of the alarm. When the siren went off (and sometimes I set it to go off even when I’m not infusing — for training purposes) I’d toss him a very desirable treat.

After a few days of doing this, when the alarm sounded, I’d watch for any behavior I could click before tossing the treat — the flick of an ear, opening his eyes, looking up. Any movement that said, “Hey! Where’s my treat?”

I c/t his reaction because now he’s offered a behavior I can shape. Over time, I’ll wait to click till he lifts his head higher, sits up, moves toward me or the pump, etc., until he is running over, demanding his treat.

The ultimate goal is for him to wake me up when the alarm goes off and keep bugging me until he gets his treat. This will work best if he keeps offering behaviors (nudging me, jumping on the bed, barking at me), until he gets what he wants.

Almost everyone agrees that a happy, effective assistance dog has to want to work — in essence, that the dog is choosing the life of a SD.

When I see Barnum eager to offer behaviors, it shows me that he has the desire to keep training and learning. This gives me great hope and optimism at this crossroads in our training journey.

By training this way, are Barnum and I meeting Patricia McConnell’s definition of a decision — that of cutting off all other options? At first glance, it might seem we’re doing the opposite: I’m allowing (encouraging, actually) Barnum to try out many different behaviors.

However, along the way, yes, I am eliminating the choices I don’t want. When I c/t for “scratch on the sandpaper,” I’m closing the door on “scratch on the plain wood surface of the board.” Of course, Barnum can still choose to do that, but he doesn’t get anything out of it, so he makes the choice I want instead.

In the larger sense, the biggest picture of all for me right now, is that breakthroughs like we had in our first “twirling” session have set me more firmly on the road to committing to Barnum as my future SD team. For a long time, I have referred to Barnum as “hopefully my future SD” or “a potential SDiT” or “a SDiT candidate.”

I do think that many people apply the label “service-dog-in-training” too soon. How can an eight-week-old puppy be a SDiT? There is too much unknown. A puppy is a puppy, learning puppy things, like where he should pee and poop and what is okay to chew and what is not. I wanted to wait until I committed to this label.

It is a very individual decision when to switch from calling a dog a “SDiT candidate” and when to say he is a SDiT. I decided to hold off on this judgment to see if he seems to have the drive to work — the creativity, if you will — before I graduated us to “SDiT team.”

Now Barnum is starting to show more and more of the traits I look for in a service dog. I have eliminated the option of seeing him as anything other than my future service dog. I have made the decision that he is my service-dog-in-training.

All this from a little good-for-nothing trick-training session. Are you listening, Universe?

-Sharon, the muse of Gadget, and Barnum (SDiT!)

Please show your love for the Assistance Dog Blog Carnival and read the many other fabulous posts. Also, let the bloggers know you’re reading! We all love comments!

http://www.youtube.com/watch?v=s7freBkTIbY

[1.] Clicker trainers use three other methods besides shaping to “get behavior” (have the dog do something we can reinforce). We can either lure (such as moving a treat up from her nose so that she follows it into a sit), capture (wait for the dog to sit on her own and click and treat), or use targeting (where the dog follows an object, such as your hand, into the position you want). Back to post.