Part 1: "Why I don't like 101 TTDWAB"
"101 Things to Do With a Box" is potentially a very stressful game for dogs who are not comfortable with shaping. It is also confusing. It can be useful for advanced clicker dogs, but I do not recommend it for crossover or beginner dogs.
The way the game is played according to Karen Pryor (who invented it), the dog is clicked only if she offers a NEW behavior with respect to the box (or whatever other object is placed before her). Say you put a cardboard box in front of the dog -- she looks at it, you click. She looks at it again -- you don't click. She might go sniff it. That's new, so you click. Maybe she looks at it -- that is not new, so you don't click. She sniffs it again. Also not new, so no click. Now what? A dog who is ready to offer a lot of behaviors may poke it with her nose (click, it's new) or tap it with her right paw (click, it's new). A dog who is not ready to offer behaviors will sit or lie down -- because that's always been safe! She will stop interacting with the box and may get very frustrated.
Now, what did she learn? She learned: "If there is a click, do NOT do whatever I just did again. It doesn't work. The only predictable thing in that little session was that if I got clicked for it once, I will NOT get clicked for it again."
But wait! What do we want our dogs to learn that the click means? We want them to understand the click to mean: "That's right, and you get a treat." "Do that again, and you will be rewarded again."
This is the OPPOSITE of the lesson they learned with 101 Things. How confusing is that? No wonder the poor dog is stressed: On top of not being accustomed to offering behaviors in the first place -- we just changed the rules.
*Some* dogs have the confidence to just start whacking the box and can figure out the totally different rule structure. This can be good -- we've got them offering. But then we go back to shaping a pre-selected task, say, "back up," the following day. Our dog backs up two steps, we click, and she may think "OK, then, no more backing up for me! They can't fool me -- I know the rules!" Uh-oh.
Advanced clicker dogs can figure out that there is a difference between 101 Things sessions and regular shaping sessions. But the OP here does not have an advanced clicker dog. She has a beginning clicker dog who is ALREADY shutting down because she doesn't understand what to do and is getting a low rate of reinforcement.
Part 2: Shaping 101
Instead of 101 Things, I much prefer to shape a single behavior with successive approximations. I choose the behavior based on it lending itself readily to the dog offering a useable first approximation readily. There are two procedural choices here:
1. Click the first thing the dog does, and then build on it.
You can use a box if you want. Or a target on the floor, or your hand, or an empty food dish, or anything else your dog is likely to investigate. If the dog looks at it, click. If the dog looks again, click again.
When the dog is pointedly and obviously LOOKING (hello, where is my click and treat???), then you will raise your criteria just a notch. Say you make it a slightly longer look. Or say you wait for the dog to now take one step toward the object. You click that until the dog is into "Hello, where's my click, I took the step!" and then you again raise criteria -- two steps, for example.
Many dogs will immediately go sniff something that is out of place, so you can get a whole bunch of behavior all at once. Click, let her come get the treat from your hand, and see if she goes to sniff again. If not, then click for looking. But the point is, when you change criteria, you're NOT saying "now do something totally different." You're waiting until the dog is confident about what to do and then making it just a tiny bit different, but still mostly the same. That is shaping and it's much less stressful than figuring out "I'm supposed to come up with something different."
In this exercise, you follow along with the dog's natural impulses for interaction. If you have a sniffy dog, she will sniff the target, and you can then build through sniff, nudge with nose, push with nose, push further with nose. Or through sniff, nudge with nose, push with open mouth, pick up (you can shape a retrieve like this). If your dog is footy and first offers pawing of the object, then you might raise the criteria through tap with foot, tap harder with foot, drag paw on object, tip object a bit, tip object over with foot, etc. You're always building on the previous step, but you let the dog lead -- if you envisioned nose pokes but she'd rather whack with a paw, fine, you build up paw whacking. She's still learning to offer a bit more when a click is not forthcoming.
2. Pre-select a behavior and build it.
In this one, you lead -- the dog does not. If I were choosing this option, I'd make sure the setup was more likely to entice the dog in the direction I wanted her to take. For example, if I wanted to shape the start of a retrieve, I might use an object the dog has in fact carried in her mouth before, or I might smear some peanut butter on the object so she's more likely to open her mouth while interacting with it. If the dog isn't getting what you had hoped was a hint, this exercise may force you to REALLY split criteria, so as to keep the rate of reinforcement high enough to keep the dog in the game. If she looks at the object, but doesn't offer a first step, then you may need to click a longer look, then a slight weight shift toward the front. This is GREAT for the trainer, by the way.
Sometimes I will prompt movement by walking around. In flyball, I often have to shape a paw target, and many dogs, if you stand there, will also just stand there and, well, not move their feet! So I may walk around and click for their foot movement. In the beginning, I click for ANY foot movement. Then I try to stop moving myself, to fade that as a cue and distraction.
My personal first shaped behavior for most beginning shapers is to target my hand. Most dogs will automatically sniff an offered palm or fist. I click this and then re-offer my hand just two inches or so from their nose. Most dogs will sniff again and I can gradually build some distance, then change horizontal angle, then height. Then I add a lot of distance. For shy dogs who are uncomfortable with the social pressure of a hand in their face, having them target an interesting object across the room will go much better. (Throw the food away from the target and away from you to get them moving around after each click.)
A next shaped behavior, and a very useful one, is "go to your mat." This is another targeting exercise. We click for the dog checking out, then stepping on, then putting more feet on, the mat. Then we raise criteria to add a little duration, and then see if the dog will offer a sit or down once he goes to the mat. This can be built into a lovely relaxing down stay that is great for helping control door greetings, dogs underfoot in the kitchen, and dogs too rambunctious around toddlers.
A last note: Some dogs who shut down during initial shaping are shutting down because of either social pressure or because they dislike the sound of the clicker. If they are starting to avoid the clicker or are flinching, I'd ditch the clicker and use a quiet voice marker to teach initial shaping. I might reintroduce a neutral sound (a quieter clicker such as the iClick, a ball point pen click, etc.) when the dog is more confident with the process. If the problem is social pressure, this often shows up as a dog who seems to respond nicely to a few clicks and then goes and sits down and sort of freezes up. For this dog, (a) note that you have a shyness issue that is a bigger picture issue than the clicker shaping one, and (b) do your shaping three clicks at a time! (Yes, animals can learn in sessions of three clicks.)
Part 3: Overcoming lure reliance and "boredom"
I doubt this is boredom. It's more likely that she has come to believe that if there's no treat in your hand, then (a) it's a different behavior you're looking for (different cue, the cookie is part of the cue), and/or (b) she thinks the rate of reinforcement is lower when you have no treat in your hand. Since scent itself can be reinforcing, she is probably right.
With a dog who doesn't do X without a lure, here is my solution: teach her that the exact opposite rule now applies. You ONLY get the reinforcement when I am NOT holding a treat in my hand. Start this with something easier, like sit.
Repeat a bunch of times. Then:
Do this twice, then go back to the first exercise for several repetitions. We want her to get the contrast: Food happens with NO treat in the hand! NO food if there IS a treat in the hand!
With heeling, once she gets the idea with sit, heel a few steps with a treat in your hand. Praise warmly. Then remove the treat, show her the empty hand, and heel a few steps with your hand in *exactly* the same position you would use if you had a treat in your hand. (The *only* change is the actual presence/absence of the food.) If she stays in heel for a couple steps, click and feed from the bowl a few feet away. From there, build distance.
I think from your description she is a very soft and worried dog. I suspect "practically goes to sleep" is her way of checking out of a stressful situation and she is actually not that sleepy when she looks this way.
Since she is a Vizsla (er, I'm basing this on your email ID -- if she's not a Vizsla, then ignore the following generalization!), chances are she is very easily worried by any anxiety on your part. This is a breed where even an otherwise cheerful, resilient dog can easily become depressed and anxious if MOM gets a fleeting expression of frustration on her face.
Unless she is experiencing social pressure or fear of the clicker sound (see my other post), I think the answer here is to keep the rate of reinforcement very, very high. Figure out how to click at least 30-40 times a minute. This means breaking out really tiny increments of behavior, i.e. splitting it very finely. It can really help to have someone experienced with shaping give you a hand at first.
One sort of unusual idea you might try will yield a low rate of reinforcement, but involves a largely involuntary behavior that she will do whether she knows what the click is for or not. The bonus is that it will relax her physically. This is clicking for blinking. If she tends to lie down and give up... she is very likely to blink from time to time. Keep clicking that until she is blinking more often! (This can take many sessions, depending on the dog.) This physiologically feeds back to a more relaxed state, which will actually make her feel better. Sounds odd, but it should work. Don't mix this with anything else in a shaping session for now. It will confuse her.
There are many,
many dogs who have been lured all their lives and have no trouble whatever
learning to be shaped. I'm not sure why it's so hard for some. Dogs
who have actually been punished for offering behavior... OK, I understand
why that would inhibit more offering. But dogs who have not been punished,
but have simply been lured, can go both ways. I am sure there's a big
personality element here -- I'm sure it has a lot to do with how the
dog tends to handle stress. Dogs who shut down when stressed obviously
are going to have more trouble with shaping!
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