Click! Good Trainer!
Last time, I admitted that I hadn’t done any of the counter-conditioning that the veterinary behaviorist and I agreed I should do with Maisy. Thankfully, no one can shame me this time, because I’ve been working diligently for the last two weeks.
I chose to focus on the noises the cats make because Maisy seems to think that her official job title is “Kitty Cop.” She takes it so seriously that she will often jump while napping in order to do some police work. Eventually, I hope to decrease some of her resource guarding with the cats, too, but I decided to start with the easier stuff first.
The first few days, I used the clicker, mostly because it is such a strongly conditioned secondary reinforcer that it can pierce her consciousness when my voice can’t. The process went like this: a cat would do something “naughty” (for example, scratching the cat tree), and I clicked. I tried to click before she dive-bombed the cat, not after she’d already started moving towards them. If I wasn’t fast enough, I simply didn’t click. I don’t want her to think that I want her to rush at them, after all! This wasn’t easy- there was only a split second in which to mark the behavior I wanted- but I was successful most of the time.
Once she figured out that staying near me resulted in treats, even when the kitties were naughty, I switched to a verbal marker/praise. Not only is it easier to use (after all, I don’t just sit around with a clicker in my hand), it is less arousing, and (I think) less reinforcing if my timing is off.
The hardest part right now is to catch as many opportunities as possible (sometimes I’m distracted, or napping, or otherwise engaged in life). As a result, my counter-conditioning efforts aren’t as consistent as I’d like, and our progress is a bit slow. Even so, I would say that about half the time she’s either looking at me instead of dive-bombing the cat, or she’s interrupting the dive-bomb to come to me. I’m pretty happy about this!
The Dumbbell
I’ve continued to work with Maisy and her dumbbell. Shortly after I posted about it last, I received an email from Canis Clicker Training about teaching the retrieve. How fortuitous! They agreed with all of you who said that the hardest part is teaching the duration on the hold. According to them, the trick is to click when the dog’s teeth is around the dumbbell, and not when the dog is curling his tongue to spit it out. However, since it’s difficult to see inside the dog’s mouth, they recommended gently pulling on the dumbbell and clicking when the dog grips on to it in order to get a nice, firm hold.
So that’s what I’ve been doing:
I’ve been starting off each training session with a simple grab. Then I move on to lightly tugging the dumbbell, gradually making the tugging a bit firmer, and with a bit more duration. I was actually surprised by how quickly she grasped the concept- it only took a day or two until she seemed to understand that she needed to hold onto the dumbbell. After she got that, I started letting go of the dumbbell for just a split second before taking it and gently tugging again. This took a bit longer- she seemed to think that she should give it to me when I reached for it- but again, she’s figured it out.
In this session, I’m working on moving my hand successively further away before grabbing the dumbbell again. This adds both distraction and duration to the hold. You can see that when I moved my hand too far away, she failed. I’ve been trying not to let her fail more than twice in a row before making the task easier again. (She actually did great in the session after this- no failures at all!- but of course, I didn’t have the video running then.)
I’m glad I took video of this session. I’m pleased with my rate of reinforcement. Even though she is distracted by the door mysteriously opening about 30 seconds in (cat, I assume), she got seven clicks and treats in 60 seconds. It also made me question if I’m holding the dumbbell too high. I guess it doesn’t matter much- I’m hoping to have her pick the dumbbell up off the ground soon- but it might have been affecting our progress. Any thoughts?
Miscellaneous
I’ve been kind of screwing around with free shaping lately, just for fun. One day, I started out trying to shape a spin, but instead ended up with her sidestepping to the right. That’s kind of neat, so I think I’ll keep playing with it.
I’ve also been thinking a lot about how my body affects Maisy. Both of us are highly dependent on body positioning and movement- shoulder movements for heeling, leaning backwards on fronts, etc. I’m trying to figure out how much of this is desirable and helpful, and how much of it is impeding our progress. Certainly, if we ever return to the competition ring, it will be nice to have subtle, legal ways to cue her. However, the trade off seems to be that she pays very little attention to my words. Sometimes, I get the feeling that’s she’s guessing, even on the basics like sit or down. More on this soon…
Showing posts with label physical cues/hand signals. Show all posts
Showing posts with label physical cues/hand signals. Show all posts
Tuesday, December 28, 2010
Tuesday, December 21, 2010
Ian Dunbar Seminar: Lure-Reward Training Done Correctly
All of Maisy’s early training was done with lure-reward. As a novice owner, it was easy and intuitive, and Maisy learned quickly. I absolutely agree with Ian that lure-reward is the best method for teaching novice owners who just want the basics. However, lure-reward training is not without its critics, and I remember seeing people in my puppy class that struggled to get their dogs to work without food. Today, I’m going to share what Ian said about using lure-reward training correctly.
The first step is to phase out the lure as quickly as possible. The lure should be used briefly, as an instruction to do the dog, and you shouldn’t use a food lure more than six to twelve times before you move to luring with air-cookies. After another six to twelve reps, your air-cookies become a hand signal! Once you’ve reached this step, the treat should always come out of the bag/your pocket/wherever after the dog performs the behavior. Otherwise, you run the (very high) risk that the treat becomes a bribe, not a reward.
Next, you need to get the behavior under verbal command. Since lure-reward training so effective at getting behavior, Ian says that you can name it right away. However, he emphasizes standing absolutely still when you say the cue so that you don’t introduce any inadvertent body cues. After you’ve given the command, wait for half a second before you lure/use air cookies/give your hand signal. If you say the word and lure/give a hand signal at the same time, the dog will be focused on the food/your movement, not your voice, and as a result, is unlikely to pick up that what you said was a cue. When your dog responds to the word, and not your physical movements, have a treat party!
Now you need to motivate your dog to want to work for you. Treats are awesome, of course, but you need to phase them out, too. Ian does this in a number of ways. He asks for more than one behavior before giving the treat. He uses differential reinforcement to reward only the best responses (he recommends using a 1 to 3 ratio as ideal). And most importantly… use life rewards!
There are three main kinds of life rewards: playing with other dogs, walking and sniffing, and interactive games like fetch and tug. Ian said that if you have a dog that plays, you have a trained dog because it is so easy to use a toy as both a lure and a reward. (Incidentally, while he does phase out treats, he doesn’t phase out interactive games ever.) Any time your dog wants one of these life reward activities, ask for a behavior first. When he complies, release him to go play or sniff or chase the ball. Then have training interludes frequently (every 15 to 30 seconds of play, or every 25 yards or so when on a walk) to request a behavior before letting the dog return to the activity.
Finally, you need to figure out how to force a dog to do what you want… without coercion. This is where RRNR comes in, and since I’ve already written a whole post about it, I won’t belabor the point.
Following these suggestions will prevent lure-reward training from turning into bribery. My puppy classes with Maisy were very good about getting us to phase out both the lures and the treats, but didn’t emphasize verbal control- we used a lot of hand signals in class, and Maisy and I are still quite dependent on them to this day. We talked a little bit about using life rewards, but I wish we would have practiced the concept to help drill it home.
What about you guys? Did you use lure-reward to teach your dog the basics? How did you do at phasing out treats, getting verbal commands vs. hand signals, and motivating your dog to work? I’m curious to hear about your experiences!
The first step is to phase out the lure as quickly as possible. The lure should be used briefly, as an instruction to do the dog, and you shouldn’t use a food lure more than six to twelve times before you move to luring with air-cookies. After another six to twelve reps, your air-cookies become a hand signal! Once you’ve reached this step, the treat should always come out of the bag/your pocket/wherever after the dog performs the behavior. Otherwise, you run the (very high) risk that the treat becomes a bribe, not a reward.
Next, you need to get the behavior under verbal command. Since lure-reward training so effective at getting behavior, Ian says that you can name it right away. However, he emphasizes standing absolutely still when you say the cue so that you don’t introduce any inadvertent body cues. After you’ve given the command, wait for half a second before you lure/use air cookies/give your hand signal. If you say the word and lure/give a hand signal at the same time, the dog will be focused on the food/your movement, not your voice, and as a result, is unlikely to pick up that what you said was a cue. When your dog responds to the word, and not your physical movements, have a treat party!
Now you need to motivate your dog to want to work for you. Treats are awesome, of course, but you need to phase them out, too. Ian does this in a number of ways. He asks for more than one behavior before giving the treat. He uses differential reinforcement to reward only the best responses (he recommends using a 1 to 3 ratio as ideal). And most importantly… use life rewards!
There are three main kinds of life rewards: playing with other dogs, walking and sniffing, and interactive games like fetch and tug. Ian said that if you have a dog that plays, you have a trained dog because it is so easy to use a toy as both a lure and a reward. (Incidentally, while he does phase out treats, he doesn’t phase out interactive games ever.) Any time your dog wants one of these life reward activities, ask for a behavior first. When he complies, release him to go play or sniff or chase the ball. Then have training interludes frequently (every 15 to 30 seconds of play, or every 25 yards or so when on a walk) to request a behavior before letting the dog return to the activity.
Finally, you need to figure out how to force a dog to do what you want… without coercion. This is where RRNR comes in, and since I’ve already written a whole post about it, I won’t belabor the point.
Following these suggestions will prevent lure-reward training from turning into bribery. My puppy classes with Maisy were very good about getting us to phase out both the lures and the treats, but didn’t emphasize verbal control- we used a lot of hand signals in class, and Maisy and I are still quite dependent on them to this day. We talked a little bit about using life rewards, but I wish we would have practiced the concept to help drill it home.
What about you guys? Did you use lure-reward to teach your dog the basics? How did you do at phasing out treats, getting verbal commands vs. hand signals, and motivating your dog to work? I’m curious to hear about your experiences!
Subscribe to:
Posts (Atom)