ClickerSolutions Training Articles

Stimulus Control

Stimulus control means that not only does the dog do the behavior on cue, but she doesn't do some other behavior when you give the cue, and doesn't do the behavior in question in response to a different cue.

That means, in a training situation, she does "down" ONLY when you cue down, not when you cue "sit" or "stand," and that she downs, and only downs, when you cue "down," the first time.

It's very common for a positively-trained dog who hasn't yet reached true stimulus control to "mix up" what should be "known" cues.

In traditional training, the dog is being punished for wrong answers, and so is motivated to avoid wrong answers. But additionally, the dog IS receiving information which allows her to distinguish the wrong answers. So stim control tends not to be a problem.

In positive-reinforcement training, we are relying on extinction (lack of reinforcement for wrong answers) to do that job for us, and it's a whole lot easier for incorrect responses to hang on a lot longer since the dog may be willing to keep offering the wrong response if it pays off even infrequently.

How come it pays off at all? Because we aren't really paying attention to what we are reinforcing.

If I teach "sit" first, and then teach "down," my dog is likely to offer a lot of sits during the training process. I may reinforce "sit" even if I cued "down," on the theory that the dog is getting frustrated, or because I just slipped up. Now the dog is getting reinforcement for both sit and down when cued "down." That's confusing.

Further, even if I don't reinforce the sit in response to a "down" cue, I might re-cue down. Now the dog figures out, "guess 'sit' wasn't the right choice... must be that other thing," lies down, and gets rewarded. So basically there's no real percentage in learning the right cue, because you will always end up giving her the answer.

By the way, this is one reason NEVER to add a cue until the behavior is really solid -- your dog may end up not really knowing what it means. Until you are systematic about stim control, your dog is going to be confused sometimes.

How do you fix it?

  • Once the dog is getting it right (offering it) 80% of the time, you must COMPLETELY stop rewarding unless the dog does it right on the FIRST try. I might give a second cue 1-3 times while training stim control, but if it happens more often than that, then I know I've moved ahead too fast and I will back up.
  • Start cuing the behavior you're working on, reward when the dog complies, then re-cue the same behavior. If your dog does it again, or stays put (depends on the shape of the behavior), reward. If the dog changes what she is doing, she still thinks toggling (guessing) is the way to get it right. Do not reward. Example with sit vs. down: "Sit" - dog sits. C/t. "Sit," and before dog has a chance to move, c/t. Catch her before she downs! Repeat this quite a few times. Then, "sit," c/t, "sit," and just wait. If the dog downs, no cookie, wait at least 5 seconds before resetting. (The contrast must be clear - there really is no reward for that incorrect response.) If she stays sitting, c/t! She is starting to understand the concept of the sit position -- it's not an action, it's a position, and if she's already in that position, then holding still is the right answer. Repeat with down, and then stand.
  • Cue a behavior, then say a nonsense word. "Sit" - dog sits, c/t. "Blue," dog stays in sit, c/t; or, dog downs, no c/t.

If you are teaching sit and down using luring, which I am totally OK with for most dogs, teach them separately. I would teach down first if possible, and lure it from a STAND. Then put it on cue. THEN teach sit and put it on cue. If you have already taught sit, reteach down by luring from a stand. This makes a huge difference and helps the dog understand that down is NOT "the thing you do after a sit if you don't get a cookie for sitting."

For "go to mat" and "go to crate," back up and make it very easy for her to get the cued behavior right. If you are working the mat, put the crate ten feet away and stand next to the mat. Reward a bunch of times for getting it right. If she goes to the crate, just shrug and "turn off" (like someone pulled your plug) for about 10 seconds. Then in another session, switch the positions so the crate is near you, and work the crate cue. Then in another session, put them about equidistant. Reward ONLY if she gets it right the FIRST time. If she is getting 80% or better, you can go on to more difficult choices (e.g. you're next to the mat, you send to the distant crate). If she is getting 60-80%, make the correct choice the easier one again for a while.

This should fix it.

Greta Kaplan
nickelsmum @ yahoo.com
copyright 2015 Greta Kaplan

 

| Training Articles Contents || Site Home |


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, malexander @ clickersolutions.com