On Saturday, I attended two lectures on training. It has been a while since I last sat and listened to a lecture on dogs!
We are attending a training course with Una, and the lectures started it. I have not been training Una much, yet. I asked the trainer which exercises would benefit the course, and will train and practice those with Una before we attend the training sessions. The idea is to teach some tricks to Una very well, in a relaxed state of mind. These tricks can then be used to change her mood in stressful situations. It’s a very handy tool.
The first thing you teach to an animal will be the strongest throughout her life. If she doesn’t know what is expected of her, she will offer this. With Una, it is sitting, followed, very closely, by offering her paw. Sitting is, obviously, one of our chosen tricks. And it is a good one for the intended use. Another useful one is targeting: touching an object with her nose. This is my second choice for Una. A third one is looking at me (or going further, making eye contact) when I say “Una.” My last choice is, probably, coming to my side when I pat my thigh with my hand.
The first step was pairing a sound with a treat so it will be easier to mark the exact moment Una does what I want from her. This is a good tool when you need to teach something new, makes the work more accurate. I chose to use a click of my tongue as the sound. The first step was giving Una treats, and making the sound every time she took the treat into her mouth: classical conditioning (click=good taste in my mouth). I did not include a video of this, because it can be explained very accurately.
Next, I chose a task Una needed to perform to get the treat: targeting. I did not plan this very well, being lazy, and just started doing things to see how it goes. First, I took one of my cats’ toys, a small fluffy ball. Una did not like it, though, and I chose to use my finger. It’s handy in a way, since it’s always with you.
I should have started shaping the behavior from a very small turn of Una’s head towards my extended finger, a thing that happens, by accident, very frequently. I decided to cut a corner and touch her nose with my finger instead. Touch + click -> Una got a treat. I used a lot of repetitions on this phase. Next, I started to wait and see what happens when I extend my finger close to Una’s nose. See the video (and notice the assisting trainer making observations from the sofa):
(I just watched the video. I don’t know how it is possible, but the sound is off sync with the video! It sounds like I click well before Una touches my finger. When you look at my face, you can SEE when I click with my tongue! I need to see if I can fix this somehow, very annoying, grr… But until I find a solution, it’s best to turn the sound off while watching)
This is a good example of the power of clicker training, and the effect of chance events when an animal gets frustrated. Touching my finger with her nose was the intended end result. When I started to wait and see what happens when I extend my finger, Una touched my finger a few times with her nose, often by accident. When she happened to use her teeth, I decided to reward for this, too. So, the end result for our session was Una “biting” my finger. It was hard not to laugh when I saw where we were heading 😀
Of course, I could have extinguished this behavior by not rewarding for it after a few repetitions. Since I plan to introduce taking toys into her mouth, I decided to reinforce using her mouth when she happened to start doing that. I hope I won’t have to live to regret this – I can just see Una and me in a stressful situation, looking at another dog, Una vigorously targeting my finger with her teeth. An absolutely hilarious prospect! Well, as you can see on the video, I actually stopped the exercise after a few bite-touches in a row. I did this simply because we had used all the treats… So, Una may or may not offer this behavior when we start the next session, we’ll see 🙂
On the video, you can also see the phenomenon I described above. It is evident especially in the beginning, where you can see me just sitting and trying to decide what to do next. And towards the end, where I start to just wait until Una figures out what to do to get rewarded. Whenever Una doesn’t know what I want from her, she tries to offer her paw and to lie down, and even to roll on her side. I am just removing my finger when she does that, removing the possibility to get a reward. I could also leave my finger where it is. I did this 1-2 times, but then I got this: Una offering her paw and touching my finger simultaneously. And this is not what I want to reinforce. Removing the opportunity to touch while fussing with her paws worked well: Una started to think other ways to get the reward 🙂
I want to make one more point on the video: my reinforcement rate is not frequent enough, most of the time. In plain words, this means that the time between two treats is too long. What does it matter? If an animal is figuring out what she needs to do, I need to reward often enough for her efforts. Otherwise she might give up. The main reason for low reinforcement frequency: I expect too much. I should reward for smaller changes to the right direction, and would be able to reward more often. On the video you can see that Una isn’t giving up, she is thinking hard the whole time. That’s why I allowed myself to cut corners and wait for a bigger effort to the right direction. A professional trainer would not do this. But we are just having fun, and as long as it works well enough, we’ll settle for the less than perfect end results 🙂
It was fun! And a very active way for Una to earn her dinner: she ate her dry food as treats for doing tricks 🙂