Once upon a time, there was a girl who decided to teach her dog some tricks. She figured out that if she gave her dog something he liked after he did something she liked, he was liable to do the thing again. So she taught him some simple tricks using food and play as reinforcement.
As she went along, her dog started finding playing training games lots of fun in and of themselves. But she still used food and play. He liked earning his “pay” and she liked giving it to him. She didn’t see any reason to stop.
This girl was unusual in that she didn’t try to tell her dog what to do in words. She realized what is not obvious to so many of us: he didn’t speak English. Things worked out just fine because he could generally discern from context and her gestures what she wanted to work on.
She used a little platform to teach him to pivot in a circle. He would put his front feet on the platform and walk around with his back feet and rotate. He got good at this and soon could spin in both directions. As soon as he saw the platform he would run over to it and start to pivot, although she could ask him to stop with a hand signal.
Some of my most popular posts are about common training errors. It seems that I have an infinite supply, and I’m willing to use myself as a naughty example. New errors keep popping into my consciousness (and my training) all the time.
In this post I’m going to focus on two main categories of errors: problems with criteria, and problems with food handling. Can you identify with any of these? Continue reading “6 Common Dog Training Errors”
Can you guess which dog got accidentally punished during a shaping session?
I charged straight into a positive punishment scenario by accident recently.
I’ve been somewhat in the training doldrums lately, probably because I am putting so much energy into finishing my book. I have several training activities that I fall back on when I don’t have much energy. They are fairly unchallenging for me (read: I can’t mess them up too badly) but still fun for the dogs. Even those have been hard to do lately.
But the other day I had some time and energy and decided to play a shaping game with each dog. We don’t shape that often, but they all enjoy it, and it gives their minds a good challenge.
Two of the dogs, Summer and Zani, used to have a default backing up behavior that they would offer in excess whenever I set up shaping games. This was my fault: my shaping setup resembled my backing up setup too much. They had similar setting factors, if you want to get technical about it. I managed to get the dogs unstuck a while back with some carefully arranged object interaction sessions. (It’s easier to get the idea to go forward instead of backward if there is something to go forward **to**.)
So I decided to set up an object interaction session again. I set out a target stick, a plastic lidded box, and a laundry basket with a plastic dumbbell in it. I put the plastic box on top of one of their mats so it wouldn’t skid around. I had an idea of a behavior for each dog, but was also willing to decide on the fly if someone did something unexpected.
Clara’s Shaping Session
Clara’s session was easy. The dumbbell in the laundry basket was for her: she loves to pick things up. I stood about 15 feet from the array of objects and it took Clara only a few clicks/treats to get over there. I stayed at a distance. This helps the dogs learn that the reinforcement zone is not always right on me, and also sets me up to practice my treat tossing.
It didn’t take much to get Clara over to the basket and looking in. The basket wanted to skid on the floor so I did go over there and brace it, at which time Clara was happy to put her two front feet in. Then I shaped her into picking up the dumbbell. This is normally very easy, but it was a slight challenge with her front end in the basket and back end out. A good time was had by all.
Summer’s Shaping Session
Summer is my super-duper shaper, which is interesting since she is my crossover dog. Crossover dogs are often reluctant to offer behaviors, but what can I say? Summer got over it. And turned out to be a creative genius when it came to thinking up stuff to do. But this session didn’t require a virtuoso performance. I shaped her to go to the target stick, which I had put behind the other stuff. She still got to it in a minimum of clicks. I needed to make more of a challenge, so I put the stick in the laundry basket with the dumbbell. She needed to hop in the basket to nudge the stick, and that she did. She’s great about getting in things.
Zani’s Shaping Session: Punishment Happened!
You knew it would be Zani, right? My easy dog/problem child.
So I had intended the plastic box for her. She’s done quite a bit of perch work and enjoys it. No big deal. We messed around a bit: she investigated the target stick and the laundry basket. Finally she noticed the box. She was directly facing me, with the box between us, and put her two front feet up on it. Yay! Click, toss the treat. Then she got on again! Ditto. On the third time, I had a sudden thought to treat in position rather than tossing the treat to reset her. So as she was placing her feet on the box, I charged right over there straight at her. She’s my pressure sensitive dog. She backed off the box in shock and scooted backwards, though she did collect the treat I had hastily thrown.
I retreated back to my area, but would she approach the box again? Nooooo. So I quickly went back to rewarding other behaviors. About 15 treats later, she was willing to go to the box again. I didn’t charge at her. It took about 5 more treats to get her putting her feet on the box.
Wanta Play Behavior Analysis?
Four processes of operant conditioning
OK, here we go. We could do at least two different analyses, because not only did a behavior decrease with positive punishment, but a behavior increased/maintained as a result of my aversive high speed approach too! I’ll leave that one as an exercise for interested commenters. Let’s go over the punishment.
We always start with the behavior that changed. What was it? Zani putting her feet on the box. Increase or decrease? Decrease. Can we identify why? Pretty sure it was my running full tilt at her. Why did she put her feet on the box in the first place? We were having a shaping session and there was a box there. So the ABC looks like this:
A. Antecedent: There’s a lidded box on the floor
B. Behavior: Zani puts her two front feet on top of the box
C. Consequence: Eileen abruptly runs straight at her
Prediction: Zani putting her feet on the box will decrease
Did the behavior decrease? Oh yeah it did! Zani loves to get on things and has been reinforced plenty for it. She had just gotten in the groove of offering “box” behaviors but stopped offering them after I charged at her and didn’t interact with the box again for quite a while. That’s a decrease. There was also a decrease in her behavior in general. She got tentative and ever so slightly shut down after my barging into her space.
Why is it called “positive” punishment? Remember that positive and negative in operant learning terminology refer to whether a stimulus is added or taken away. In this case Eileen charging at Zani was an added stimulus.
So How Bad Was It?
Positive punishment is the learning process that we pretty much try to avoid at all costs. So how hard should I be knocking my head against the wall?
As usual, we ask the dog, and we do this by observing her response. Did we see side effects? Referring to the list on my post “7 Effects of Punishment,” we probably got small doses of # 1, avoidance, and #4, apathy. It remains to be seen whether we will see any avoidance of me outside of training, but I could easily see her getting sensitized about my approaching her. During the session I saw a decrease in behavior from her in general, which could fit under #4. Luckily, this was only over a brief period. Zani started offering behaviors again, and then was getting back on the box willingly (i.e., no pressure from me) within about two minutes. Susan Friedman points out that when an animal has a large reinforcement history and “trust account,” the animal can typically handle life’s little unpleasantnesses well. So this probably wasn’t a horrendous tragedy.
On the other hand, I have worked very hard to pair my approaching and entering Zani’s space with good stuff since she is sensitive to body pressure. We play games where I invite her to enter my own space as well, especially when I am standing up directly facing her. That’s just hard for her, polite little dog that she is. So chalking up another “Eileen is a boorish clod and she scares me sometimes” experience was not ideal. Even just that one time may set us back just a bit in the work I do to make her comfortable with me. In other words, there is a good chance that there is some fallout of the avoidance type, though it may be subtle.
The side effects of punishment listed are generally overt behaviors. There’s also the basic issue that it can scare or hurt an animal. Whatever the animal’s behavioral response, that’s not a good thing.
So how to think about this? I don’t think being alarmist is helpful. Yes, I punished my dog, but it’s over and done with and wasn’t a tragedy (even from Zani’s point of view, I’m pretty sure, which is the one that counts). But neither do I think this is the kind of thing to brush off. It set us back just a little bit. Zani might be a little extra wary with me in certain situations for a while. I’ll have to work that much harder to make approaching her in various ways into a happy thing.
Accidentally running up in my dog’s face is not something most people would design as a deliberate punishment. People who do use positive punishment in training would probably be amused that I am even classifying it as such. But one of my points is that even such a benign-sounding action can have fallout. Why use punishment to decrease one behavior when it will simultaneously create problems with others? You are left always trying to fill a leaky jug.
And Zani, though sensitive, has a pretty solid temperament and is used to my ways. What if I had been working with a fearful dog or even one who was new to me? A mishap like this could have meant a setback of days or weeks.
Anyone want to share their own accidents? I’m not asking for true confessions about deliberate aversive use. Plenty of us have those in our histories. Let’s talk about that another time. I’m more interested in the boo-boos. I bet I’m not the only one….
I wrote in my first post that one of the things I have to offer the world is a window into my mistakes as a newbie trainer. Last week I posted an update on the smashing success of my feral dog Clara (which I can’t really take credit for; most credit goes to my teacher).
So today I’m going to show you something that didn’t work with one of my other dogs, at least not how I expected it to. I think it may be very helpful to some people, as I made a common error. I don’t enjoy these pictures and videos of my dear little Cricket showing stress signals, but perhaps publishing them can help some other people and their dogs.
Once upon a time there was a girl who decided to teach her dog some tricks. She figured out that if she gave her dog something he liked after he did something she liked, he was liable to do the thing again. So she taught him some simple tricks using food and playing with toys as reinforcement.
She and the dog had so much fun that she found as she went along that he didn’t need to be reinforced with goodies as often; he started finding playing training games with her very fun in itself. But she still used food and play, especially with new stuff or very difficult things. He liked earning his “pay” and she liked giving it to him.
It didn’t occur to her to tell the dog what to do in words, since she knew he didn’t speak English like she did. But things worked out because he could almost always discern from context and her gestures what she wanted to work on.
She had a little platform the she used to teach him to pivot in a circle. He would put his front feet on the platform and walk around with his back feet and rotate. He got very good at this and soon could spin in both directions. As soon as he saw the platform he would run over to it and start to spin, although she could ask him to stop with a hand signal.
One day she decided she’d like to teach him a new trick using the little platform. She wanted him to sit on it. She got out the platform and he ran over and immediately started spinning. She laughed and signaled for him to stop and he did.
With gestures she got him up on the platform with all four feet within a few minutes, and it was easy from there to get him to sit.
The next time they played training games with the platform, he ran over again and started to spin. But she indicated to him that she wanted him to get up on it and sit, and he soon did. Each time they trained, he spun less and sat faster, until one day he ran in and sat on the platform. She told him how smart he was and gave him a cookie.
Over the next couple of weeks she had him do lots of things on top of the platform, and didn’t ask him to spin. He would always run to the platform and sit on it to start.
Then she asked him to start spinning again. They worked on both things equally. After a little awkwardness at the beginning, he always figured out what she wanted.
One day she set out to train and got the platform out. Her dog ran in and then stood stock still next to the platform and looked straight at her. He seemed to be asking, “What are we going to do today?” She realized it would be nice for him if he knew which thing she wanted him to do that day, rather than always having to figure it out by trial and error.
She thought about it and realized she could create some way to let him know which trick she wanted to work on. She made herself a silly hat out of newspaper. From then on, every time she wanted him to get all the way on the platform, she wore the paper hat. When she wanted to work on spinning and pivots, she didn’t wear the hat.
It took only a few sessions for him to catch on, and thereafter he would immediately offer the right starting behavior depending on whether she was wearing the hat or not.
Question: What did the girl create with the hat?
Answer: A cue.
What’s the Point?
OK, I’m a little obsessed with cues. But I would really like to share my (admittedly limited) understanding with those who are newer at this than I am.
First, all sorts of things can be cues. If you don’t create a deliberate, explicit one, dogs will usually figure out what you want from contexual cues. Before the girl started using the paper hat, there were still lots of cues for the dog. But they were fluid and not systematically organized.
You might not even know what a dog’s cue actually is! Lots of times when we think the dog understands a verbal cue, they are cuing off something else entirely. Try this: put your dog in front of her crate (if you use one), point, and say, “Purple cow!” Some other time, get your dog in front of the crate, don’t point, but just look at it, and say, “Daddy long legs!” Dogs notice contextual cues brilliantly, and most will get into the crate in this situation. If you had proofed the living daylights out of your crate cue and had complete stimulus control over it, as long as those two phrases aren’t your real cues, the “proper” response would be for the dog to stand and look at you, waiting for further instruction because he knew you had spouted nonsense. But almost no one puts crate or mat behaviors on stimulus control, so most dogs who are conditioned to like their crates will leap in at the slightest hint that that might be reinforceable right now.
Conversely, think of a situation in which you always, without fail, ask your dog to sit (with or without a verbal cue). Get them in that situation and give your verbal for down, stand, or another behavior and see what happens. If you have worked very hard with your dog on the distinction between your verbal cues, your dog might do fine. But most will have a bit of hard time.
Finally, cues in training or the real world don’t have to be quick words or movements. The “Open” sign that stays lit up all day in a store window is a cue that says that you can go in the store and shop for a while. When you’re at a club, the music going on is a duration cue for people to dance. Most people stop when the music goes off. You don’t have to, but it’s more fun (reinforcing) to dance while the music is on. So a paper hat, left on, can be a cue that a certain type of training is going to happen and a certain family of behaviors will likely be reinforced.
Here is Summer in a situation where the contextual cues and something called the matching law conspire to make her fail to respond correctly to a verbal cue. (Stay tuned for Part 2 on the Matching Law.)
Having clear cues is a way to be fair to your dog. Remember, a cue is an indication that a certain behavior, set of behaviors, or behavior chain, is likely to be reinforced. Having unclear ones defeats the purpose. Help your dog by being very clear about it!
Summer targets the plate so hard that her muzzle slips upwards. And look, her mouth stays closed!
I have learned a lot in the last three months. Yes, that’s how long it has taken for me to get Summer’s target behavior where I really wanted it.
Back in September I published a post about the many ways I had messed up Summer’s target behavior. I had lived with it a long time, but it really became a problem when I tried to use a target for distance behaviors because Summer kept biting it and trying to bring it to me.
One of the first things I learned was that Summer was not the only one making correct and incorrect behaviors. I was sometimes marking the incorrect response! So I added two columns, one for my number of correct reps, and the other to express that as a percent. That made me clean up my act in a hurry, and pretty soon I was no longer marking incorrect behavior except once in a blue moon.
I also added a column for a moving average, to smooth out some of the noise in the graph and show the trends better. And just because I’m a nerd and I like that sort of thing.
Recall that one of the worst (out of six) problems we had was Summer’s biting the target, because of our retrieve work before we got target on cue.
I had picked a new hand position so as to change the picture completely for Summer, but my choice, which made it appear that I was holding a treat in my fingers, elicited even more teeth and biting from Summer at the beginning. (So ironic, since Summer is not at all a mouthy dog.)
After the first few sessions I came close to changing my hand position again because of all the teeth. But I decided to take the challenge and keep it. I really liked the touches I was getting from her.
Really, this was the hardest part and took the most time. It took a little more than a month of practicing only with my hand to get rid of the teeth. But I’m really glad I stuck with the new hand position because started getting much, much firmer and nicer touches from Summer than I ever had before.
After we got the hand touch, I tried transitioning to a target stick and that was disastrous. Bite city. The stick was a cue for the retrieve work we used to do. So I thought of an object that I could hold that she couldn’t bite. How about the back of a plastic plate?
Putting the spoon against the plate made it less tempting to bite
So we did many reps with a plastic plate with a piece of blue painter’s tape on it. A good Internet friend points out that blue tape is nicely visible to dogs. After about a month of that, I brought my target stick back out (it also has tape on it), and held it flat on the back of the plate. By making an interim step (splitting), I was able to transition her to the stick without having teeth. This was a huge step, and a good one towards my practical goal of being able to send Summer to a freestanding target stick to touch.
Where We Are Now
Over the weekend I tested Summer on hand, plate, and target stick touches and we got 100% correct! Not only that, but her touches have still nice and firm and she is eager to do it. No more drive bys for sure.
You can see from the graph that I lifted from my tracking document that there are several dips in performance; those correspond to places where I raised criteria. But even counting those dips, her overall average was 86%. Keep in mind that my goal for percentage correct before proceeding each time was 95%, not the 80% that trainers typically shoot for before moving forward. You can see in the graph that we stayed on each step longer than we would have had to if that were our goal. It worked for Summer and me because neither of us minds repetition.
This graph covers 1,012 correct repetitions. Yes, you read that right. About 1,000 reps. Let that be a lesson. Try to train it right the first time!
Final Notes on Criteria and Method
I ended up changing one criterion from my original training plan. I had specified that I wanted Summer’s mouth to be closed. But I got visually confused when I saw her approach with an open mouth, then close it just before the touch. I decided that was her business whether she wanted to leave her mouth open, as long as she touched my hand or the object with her nose/muzzle and not her teeth. This worked out for us.
I wrote in my previous post that I wanted to avoid negative punishment if possible. I did end up doing it a few times. Sometimes we would get in this loop where she would do an unacceptable touch and when she tried again, one of the undesirable behaviors would pop immediately back in. So a few times when I got a bite or felt teeth, I not only didn’t give her the treat, I pulled my hand back and paused, with a little break in the action. This was always followed by a correct response from Summer. The penalty did seem to communicate very well that I wanted touches and not bites. I probably did it fewer than 10 times in our 1,000 reps.
At the time it seemed more kind than letting her try over and over again without getting reinforced (extinction). A more skilled trainer probably wouldn’t have had to do either (and certainly wouldn’t have taken 1,000 reps!)
Notes about Future Steps
What’s left, following the Training Levels, is a foot touch (her nose to my foot), then touching a Post-it or piece of tape on the wall, with the final goal of pushing a cabinet door closed.
I don’t anticipate a problem with the foot touch, but the wall thing will be a challenge because we have done lots of wall touches with her paw. But I know how to be patient, and so does Summer.
Some final tasks will be a duration touch, mixing up Zen and target cues, and finally distinguishing target and retrieve cues. And of course I’ll need to generalize every one of these things and take them on the road.
Thanks for reading! I would love to hear more retraining stories. I’m not the only one, am I?
By the way, now that it’s done, here is the whole series in one place:
As the great trainer Bob Bailey says, training is simple but not easy. The principles are very simple and straightforward, but actually applying them in practice can be very difficult.
I’ve mentioned many times that I am not a professional trainer. But I hang out with some phenomenal ones. Plus, I am a student of life and tend to do lots of observation of myself and others. (What, you had noticed?) And I don’t mind sharing my own errors if it can help somebody along.
When you follow directions from a book, such as the Training Levels, a lot of the planning is done for you. So I get a little lazy about training plans. I can just check little boxes off in the books. (A reminder: I acquired all these errors when I was brand new at training and using a mishmash of methods. Wish I’d known about the Levels earlier.) But it’s a good idea to always have a plan, and collect data on what you are doing. Since I now need to do a unique retraining plan suited exactly to my dog and her needs, I am going to make a plan and share it, then share how well it works.
Training Plans
Training plans can be as simple or as detailed as the situation demands. For this situation, where I am trying to get rid of several superstitious behaviors that often follow my cue for hand target, I am going to make a thorough plan, and share it here.
Here are two posts about training plans, both by great trainers.
I combined parts of both of these to design the categories for my plan. I also made a record-keeping spreadsheet in Google Docs, loosely based on Melissa Alexander’s. Hers is accessible through her post above.
My Training Plan
Goal: a clean touch of Summer’s nose to my hand, followed by her generalizing that to similar touches to different objects. I want verbal cue recognition (will do tests with objects, see below). But she doesn’t have to wait for a verbal if I do the hand signal.
Description: A clean clear touch of nose to hand or object. She can be in any position that will allow her to reach the hand or object. It doesn’t have to be a hard touch, just definite touch of nose. No drivebys, and no just whiskers. No teeth, no open mouth. Minimal paw lifts. I define minimal as: her paw can lift about an inch higher than normal if she is walking or trotting to the target. Getting her mouth closed and preventing paw whacks are essential. A little leftover paw action is OK with me.
Methods: Capture the touch, then shape a firmer touch if necessary. I want to make the picture as different as possible for Summer from the very beginning, including changing the hand signal and verbal cue. I will follow the progression in Level 1 Target in the Training Levels. I will start with me seated. Use Sue Ailsby’s hand position (see “new position” above). Start off with my left hand rather than right, which I have used more often for hand targets before. I’ll drop treats rather than handing them to her (encourages mouth/hand contact) or throwing them (builds excitement).
Cue: Verbal. In the case of hand touch, presentation of hand. Cue discrimination: the ability to distinguish from Sit and Down on verbal alone. For this I will use a standalone object, since the presentation of the hand will always be more salient than the verbal. When to start with the cue: TBD.
Sessions: Up to three sessions per day of 10 treats.
Criteria for advancement: In the early stages of the hand touch, 95% or above. This is because my goal is to clean out the old superstitious behaviors. Also I have observed that Summer doesn’t mind lots of repetition. Later I will build in her looking me in the eyes before I will give the cue. This is because of her habit of staring at the food or my food hand.
Duration? Not for this project.
Distance? 15 feet to object, or about that much if chasing me.
Distractions? Maybe near the end. Put down a mat for her to go by as she goes to touch an object.
Position: Hand touch from all different directions. Object touch from different positions. I will limit to objects already in her sight, i.e., she doesn’t have to turn around to find it. However, I plan to “try it cold” by cuing a Touch when she is not expecting it and when there is an obvious object to touch.
Where: Start in my den. Do other rooms in house, back porch, back yard. Possibly go on to front porch.
Reliability: I want 95% free from superstitious behaviors. Response to cue itself 80-90%.
Comments and caveats: Since we have an ongoing issue with staring at food, I will chain in eye contact after she is getting some fluency. She is more likely to do the undesired behaviors if she is excited and moving fast, so I will start with her standing still. Observation: she is quite likely to offer an undesired behavior after failing to meet criteria and doing a light touch on the first one, instead of offering a firmer touch. I will need to be creative and use positioning to avoid errors. Also I stated earlier that I don’t want to use negative punishment at all if possible. That means I don’t want to rely on pulling the target away from her if she is approaching it with her paw or an open mouth. I want to prevent those things from happening to begin with. I want to tell her through reinforcement what is working.
Future: Duration. Mix up Zen and target. Learn to distinguish target cue from retrieve cue.
The difference between my old and new hand positions for target:
Old hand position for target
New hand position for target
Notes about Future Steps
In the Training Levels, what follows the hand touch is:
Foot touch: (Dog’s nose to human foot) Probably no problems here.
Wooden object: I’ll need to prevent teeth touches and grabbing by using a large, flat object, as described in the Levels (p 187) Need to watch for feet movement. How to discourage? Careful height of object. Experiment with stationary vs moving.
Plastic object: ditto.
Metal object: ditto.
Spot on wall: I’ll have to modify the instructions: I won’t use a post-it note or painter’s tape. (Watch the Targeting Mishaps movie to see why.) I’ll draw or paint a target on a piece of poster board with non toxic paint. Start by holding the board. Shape touching the spot. When that is solid, get it onto the wall.
We have practiced all of the above behaviors before, but many incorrectly because of superstitious behaviors.
Session Planning
Session 1. I’ll sit in a chair. Treats on my right on a desk. Proffer left hand in position described by Sue. Correct iterations marked by Yes and drop (don’t throw) treat.
Wow, real life comes crashing in. So Summer did one touch/sniff, then the very next one she took all my fingers in her mouth. (A “bite” but very inhibited. Her teeth didn’t close.) I wasn’t ready for that at all. I was in the middle of saying “Yes” but aborted it. I was so surprised I just got up and turned off one of the cameras and took a break. In the meantime Summer heard me say most of “Yes” and was sniffing around looking for her treat, which I had made a split second decision about and didn’t give her.
Dang! An important goal for me is no negative punishment, but abruptly getting up and stopping a training session can be a big dose of that….
But the video taught me a lot. Both the times (yes, it happened again) Summer took my fingers in her mouth, I had presented my hand kind of flat. Must have looked like I was handing her a treat.
Besides the position of the hand, I need to make its presentation a little clearer (I don’t need to leave it halfway out there). Make it very clear: on/off. I’m still struggling a little with the hand position; that’s part of why I am so stiff. Also I’m trying to keep my body very quiet. A couple times I was too slow and she was already moving forward when I presented my hand.
I’m really really glad I counted reps and successes. I would have overestimated our success rate otherwise.
Also, I chose to go with 10 treats rather than 10 total iterations. 10 treats means 10 correct responses, but puts no limit on incorrect responses. Sometimes not advisable at the beginning. But even looking at the video I had a hard time deciding what “counted” as an iteration or not, so I’m glad I wasn’t trying to count while training.
Third and Fourth Sessions
We have already had our third and fourth sessions, although they’re not included in the movie. Our success rate got better and went up to 10 correct out of 13 both times, which comes to 77%. I tried to loosen up a little and move in Session 4 but I immediately got an open mouth from Summer. I’ll need to continue to be very conservative since movement on my part has typically triggered mouthiness on hers. There’s always a fine line between getting the behavior and not wedding it to a certain setup. I’ll do some other things to introduce some variety.
Bob Bailey said, “What you click is what you get.” There is a lot of wisdom in this simple remark. Among other things, it emphasizes to me that we don’t always realize exactly what we are marking and reinforcing, but the animal always does. Or rather, the animal’s actions reflect it.
Since I rarely use a clicker, my version is, “What you reinforce is what you get.” This is still a challenge to keep in mind sometimes. I tend to fail at holding my criteria steady, and it shows in the overly wide range of behaviors I tend to get from my dogs. Plus, putting something on an intermittent reinforcement schedule (reinforcing it inconsistently) makes the behavior really persistent. Not a good idea to do that to a behavior you are trying to get rid of!
So let’s see what that all this looks like. I’m going to share with you all one of my bumbles. I have a video where I can show first what I reinforced purposefully (and successfully). Then I show the dog doing what I subsequently reinforced carelessly. It happened to be very close to the behavior I had been trying to fix in the first place. My dog shaped us almost back to where we started!
I wrote in my crossover story that a turning point for me was when I learned that an animal’s behavior is a map of what has been reinforced. (Punished too, now that I think of it.) You can see the changing landscape in the movie.
Letting Go of the Ball
Clara is my first truly ball crazy dog. I love it. It’s so fun to see that pure passion; how completely thrilled she is about playing ball. She loves it so much, actually, that she has a rather hard time giving it back, even though she lives for me to throw it. She loves both chasing a ball and having a ball.
Clara practicing “put it in the bowl”
I published a movie last year, Retrieving to a Container, about how I solved her problem of reluctant releases. I did this with the help of my trainer friend and also a great YouTube tutorial. I trained Clara to fetch the ball and drop it in a container instead of putting it in my hand, which was so very hard for her. (She will fetch just about anything else in the world to hand, from paperclips to poop, just not a ball. With the ball, she approaches since she really does want me to throw, but then she usually does that head dodge thing when I reach out. Just c a a n ‘ t quite give it up.)
I could have stopped everything and worked hard and gotten a ball fetch to hand, but the container thing was an elegant solution that would also build us a new foundation behavior. And it removed most of Clara’s conflict about releasing the ball.
I tried teaching my other dogs as well, and Zani took to it right away. So now I had two of them who would drop things into a container.
Zani has a knack for getting in on the fun, wherever it is. So when I would get out the rubber balls and the container, she started barging in on Clara’s game. Clara is good natured about things like that, and I’m a sucker, so now there were three of us. Zani started to pick up the ball if Clara dropped it short of the bucket. Zani would grab it, drop it in the bucket, and I would give her a treat. (Told you I’m a sucker. She even got me to feed her.)
Experienced trainers are smiling now. With Zani’s help, I exactly undid the behavior I had trained. Clara and I play with two balls, so I can throw the second immediately when she delivers the first. The throw of the ball reinforces the previous behavior. So when she started dropping the ball short of the bucket and letting Zani finish the job, she still got reinforced by another throw. It didn’t matter that I was waiting for the ball to hit the bucket, since she wasn’t performing that part of the sequence. So she reverted to her natural behavior of tossing the ball down in anticipation when she got within a few feet of me.
How Eileen’s Behavior Got Shaped
So what about me? Did Clara cause my behavior to change through reinforcement? Yes. Her actions were shaping my behavior. She got me to do two different things. First, when I was holding the container, if she dropped the ball a time or two I got in the habit of reaching out with the container before she let go. I was doing the natural human thing of “catching” the ball with the bowl, rather than being a statue. I got reinforced for doing that since it saved the time of either of us messing around trying to pick it up off the ground. So in this way I also started taking over some of what “should” have been her job, and she got reinforced (again!) for not coming quite all the way to the container. By inches this time, but it only takes that much to miss.
Take a look at my right hand
Second, she also shaped me to put the second ball out of sight when she approached. Again, she’s so ball crazy that she had a very hard time taking her eyes off the ball I was about to throw long enough to put her own ball in the container. I could have started working on her self control around balls, but instead I fell into the short cut of putting the other ball out of sight when she approached. This improved her accuracy at the container.
Where to Go From Here
All this makes me sound incredibly sloppy, but I’m going to defend myself a little. First of all, this is recreation. There are some things I put lots of energy into getting just right. Zen. Recalls. Mat work. I am even decent at being moderately precise, as in competitive obedience and Rally. So I cut myself a little slack when we are talking about something that is not life and death important. (Clara disagrees about that assessment, grin.)
Second, with multiple dogs you tend to make little compromise decisions all the time. It was a big plus in my mind that I could play with Clara and Zani at the same time, bizarre as the game was. My bottom line was for them to have a good time and me to be able to not work very hard.
However, the problem with being sloppy in any training situation is that one is changing criteria on the dog.
Changing criteria is unfair without using clear cues for the different behaviors expected. That’s what cues are for. In this situation, with a different dog from Clara, my behavior might have been more of a problem. Clara is resilient and adaptable, especially when there is a ball involved. When I firmed up my criteria it took her less than a minute to switch from dropping the ball a few feet from me back into taking some care to drop it into the bucket. But it did take a little extinction burst. I try not to get in the habit of creating those!
So in the course of filming and writing about this, I have decided how to fix this situation in a way that hopefully will be more fair to Clara than the current mishmash, and still let Zani participate. I’ve realized Clara is very close to understanding the two different criteria for when Zani is there and when she isn’t. I can do something to make it even more clear which criterion we are using. I’ll go back to sitting down when I play with her by herself. I think that change, plus Zani’s absence, would make for pretty clear situational cues that it she is in charge of getting the ball into the container.
Also, my friend Marge has challenged me to address self control for Clara around balls. So stay tuned. Finally, for extra credit: why is Zani hanging around me so close when she is part of the game?
And how about you? Have your dogs shaped your behavior? Have you noticed anything amusing that you have been reinforcing? Or noticed slippage into a different behavior as you relax criteria?
