Beginning Shaping: Planning Your Sessions
Yes, you're getting it. Believe me, the single criteria and short session concept was one of the most important things I brought back from the seminar. One of. DoG, there are so many things that's it's really hard to single things out like that. :-)
Yes, you can do multiple sessions. However, you don't *have* to change or increase your criteria each session. At the end of the session, evaluate your dog's performance. Did the dog get 80% correct? Is the behavior strong? Then you can increase criteria. If the dog is getting fewer than 80% correct, remain at the same level for the next session.
We didn't break up the sessions this much. Actually just crating her while you evaluate her progress and make the plan of where you want to go next -- literally just a minute or two -- is enough of a break. Since I have two dogs, I may either alternate dogs or do several sessions in a row with one, then several sessions in a row with the other.
I'm planning to make my sessions either 10 repetitions or three minutes, whichever comes first. I'm also keeping track of repetitions, marking the errors, so I'll be able to tell easily when the dog is ready to have the criteria increased. I'll also be able to see when I've increased the criteria too fast because my success rate will be low and will remain low for two or three sessions.
Absolutely! Have you played the timing games that the Baileys have posted on various lists? Basically, have someone drop a ball or a set of keys. You try to click so that your click and the sound of the object hitting the floor coincide. You shouldn't be able to tell them apart. Have the person drop the object from varying heights. The game gets much more difficult as the object starts closer to the floor.
Terrific breakdowns!! I'm doing the very same thing. I'm starting by describing the complete behavior in minute detail, response by response. Then I'm breaking it out into individual criteria and figuring out what needs to be taught in what order. I'm making a separte list of proofing exercises and aniticipated challenges -- what I think I or my dog will find most difficult -- so I can pay special attention to those things. Bob says this type of planning is critical but is often overlooked. Training is easy (okay, easier) once you have a very well-defined picture of what response you're looking for.
This brings up another related point. Don't be afraid *not* to use the clicker. The clicker, when used correctly, gives you incredible precision. If, however, you're working on something more general, or if you're not sure exactly what you want to reinforce, don't click -- just chuck food.
For example, loose leash walking is way less precise than heeling. So I can just feed, feed, feed while the dog is in the general position. However, if the dog is wandering all over, and I really want to indicate when he has wandered into the correct location, I might use the clicker for that. Even with loose leash walking, I like for Rain to glance up at me. So for that specific, discrete behavior, I used the clicker.
Bob says there are lots of times that they didn't choose to use the clicker in the very beginning. First they just chucked food to get some semblance of form, *then* they shaped and refined the behavior using the clicker.
Yes and no. Work on them separately. First deal with even, ignoring square. Once even is strong (80%), add square. Be prepared for even to decrease in reliability temporarily. That is okay. (Think about learning a physical skill. Every time a new aspect is added, the others go to pot while you figure the new one out, then you figure out how to coordinate everything.) Just concentrate on square. Gradually, you can require both square and even -- just give her plenty of opportunity to get it all straight in her mind (and muscle memory) before you do.
Bob says only give a jackpot when the dog vastly surpasses the criteria. Do not, however, raise the criteria to that new level. If the dog continues to offer at that new level for the rest of the session, *then* you may increase to that level in the next session. Don't change your plan mid-stream.
Well, Bob didn't really get into this. He likes food because it's a primary reinforcer, it's easy to deliver, and it doesn't slow down the session. He, I believe, uses the same food most of the time. I vaguely recall someone asking about this, and asking if it wouldn't be better to vary the food types, and he shrugged and said, "That's complicated. I like simple. Simple works."
Before the world revolts and starts talking about using a higher value reinforcer in more distracting situations, realize that Bob build an *extremely* strong, reliable, highly-reinforced behavior before he ever gets to those distractions, and those distractions are phased in, in a very controlled manner. Like a variable schedule of reinforcement being used to strengthen behavior, using a variable treat value in distracting situations is a crutch to make up for a too weak behavior.
Now, for the record, I've always made up a dry treat mix with a variety of reinforcers. I keep it in a tupperware container and just pour some into a bowl when I'm ready. That's simple for me, and my dogs like it. I'll probably continue doing that.
Also for the record, I probably will still teach variable schedules and using a variety of treats to my beginning students. Why? Because they aren't very likely to build the strong behaviors, just like Bob said. If they have interest enough to continue, or really get "into it," then I'll encourage them to switch to continuous.
That might be the reason. Another possible explanantion is that some part of your body language has become an accidental cue for that behavior. Bob warned us to keep our body language very neutral when working with the chickens. For example, if we lean forward a bit before we click, the chicken will cue off of that, not the click.
Once you add the cue -- and even before -- take pains to make sure your body is not a salient stimulus (unless it is supposed to be). Change the picture frequently. Don't just change rooms, but change your position. Lie down. Sit down. Face backwards. Stand on one foot. Bend over. Cross your legs. Hold your arms in the air. Once you add a cue especially, the only consisitent thing in the picture should be the cue.
I think you're describing a chain. Getting all the parts into place individually first, then finally adding a single cue. When Terri and Christy get back from the advanced seminar, they can give us all the wisdom of chaining.
I'd say you're definitely on the right track. Well done! Let us know how it goes.
Copyright of all posts is the property of the original author. Please obtain permission from the original author before copying, quoting, or forwarding.
List and Site Owner: Melissa Alexander, mca @ clickersolutions.com